1 Since 1.6.0002 (1.6.0b)
3 * Correct an error in mount point string construction
4 when the user provides a cell alias to "fs mkmount"
5 instead of the full cell name. Every mount point
6 string must end in a trailing dot.
8 * Major improvements to the Explorer Shell Extension.
10 * Addition of the AFS Redirector drivers and service
13 Since 1.6.0001 (1.6.0a)
15 * Correct an error in CheckOfflineVolumeState processing
17 * Permit NPLogonNotify() to exit immediately if the
18 SetToken error is KTC_NOCM.
20 Since 1.6.0000 (1.6.0)
22 * Avoid a cm_serverLock reference count leak.
24 Since 1.5.9907 (1.6pre7)
26 * Add Shutdown message to the windows event log
28 * Check Offline Volume status according to the specified policy
29 not every time the daemon thread runs.
31 * if the directory object cannot be found in the tree, return
32 CM_ERROR_PATHNOTFOUND instead of crashing.
34 * avoid memoryt leak of cm_GetSpace() allocations
36 * improve afs/kfw error message output. Include PT and KTC
37 errors. Impacts afslogon.dll, afscreds.exe and afssrvadm.exe
39 * afslogon.dll NPNotifyLogon will now kick start the
40 afsd_service if it is not already in a pending start or
43 * properly construct the userrealm in the afskfw library.
44 do not include the '@' in the realm.
46 * fix the generation of HTML reference manual pages from
49 * No longer send release lock RPCs to a file server for
50 a deleted file. Each attempt will fail with a VNOVNODE
51 abort. Multiple aborts can result in the file server
52 throttling the client.
54 * cm_BkgDaemon should not perform the job of cm_SyncOp
55 which already properly serializes operations on the
58 * cm_BkgDaemon should not perform queued tasks on deleted
59 objects. Doing so is a waste of resources.
61 * Avoid rx nat pings prior to connection attachment.
63 * After directory enumeration completes move the directory
64 status object to the most recently used position to avoid
65 it being recycled when the directory contains more objects
66 that the status cache can hold.
68 * Reduce memory fragmentation caused by lock order validation
71 * When a status object is deleted, move its object to the
72 least recently used position in the queue to promote
73 recycling when the reference count drops to zero.
75 * Fix the osi_Log macros so they safe for use in if..else
76 statements by wrapping in do..while(0). Several logged
77 messages were wrong as a result and there may have been
78 other logic errors caused by mismatched if and else
79 statements when braces were not used in the code.
81 * Apply interlocked operations to all state and queue fields
82 in the cm_call, cm_volume, cm_scache, and cm_buf objects
83 in order to ensure that operations are atomic. State flag
84 bits were being lost on multi-processor systems under heavy
87 Since 1.5.9906 (1.6pre6)
88 * add support for detection of NTFS symlinks to
89 \\AFS using GetFileInformationByHandleEx which is
90 available on Vista, Server 2008 and beyond.
92 * improve thread safety of afs_shl_ext.dll
94 * MS11-043 adds response validation for SMB_COM_NEGOTIATE
95 messages received by the SMB Redirector. OpenAFS failed
96 to properly specify a Challenge and DomainName in the
97 response when the security mode is SMB_AUTH_NONE (or share
98 with password). This patchset corrects smb_ReceiveNegotiate()
99 so that it adheres to the protocol specification.
101 * smb_T2SearchDirSingle() must not fail directory search requests
102 for the _._AFS_IOCTL_._ file. Although this file does not actually
103 exist, it is successfully processed by CreateFile operations.
104 Therefore, an explicit search for it should return a valid answer.
106 * cm_SyncOp/cm_SyncOpDone is used to synchronize the RPC processing
107 to ensure that calls which are in conflict cannot occur at the
108 same time but also to ensure that the ordering of operations
109 is consistent. cm_MergeStatus() was in many cases executed after
110 cm_SyncOpDone() removed the synchronization barrier which in turn
111 permitted status information to be applied out of order. Side
112 effects could have included data loss due to client side file
113 truncation. More commonly two StoreData RPCs would have their
114 status information applied out of order forcing the cache manager
115 to invalidate all of the cached data for the file.
117 * cm_NewServer() can result in a call to cm_UpdateVolumeLocation()
118 if a server probe is performed. In order to avoid recursive
119 calls to cm_UpdateVolumeLocation() do not probe new servers from
120 within cm_UpdateVolumeLocation().
122 * Numerous Rx improvements.
124 * Fix out of order lock acquisition in smb_ReceiveV3LockingX()
125 which occurs if the lock acquisition fails.
127 Since 1.5.9905 (1.6pre5)
128 * aklog supports dotted Kerberos v5 principal names.
130 * afskfw library always attempts afs/cell@USER-REALM
132 * afskfw library must test return code from
133 krb5_cc_start_seq_get() or will trigger a null
134 pointer exception when using Heimdal.
136 * lock protected fields must be 32-bit in order
137 to avoid memory overwrite races.
139 Since 1.5.9904 (1.6pre4)
140 * Fix caching of non-existent volumes. The test to
141 trigger an immediate CM_ERROR_NOSUCHVOLUME in
142 cm_UpdateVolumeLocation() was backwards.
144 * Prevent the background daemon from checking the
145 status of non-existent volumes. cm_CheckOfflineVolumes()
146 should skip volume groups with the CM_VOLUMEFLAG_NOEXIST
149 * The afskfw library should return an error immediately
150 if the krb5_32.dll library cannot be loaded. Affects
151 afslogon.dll and afscreds.exe.
153 * No longer depend on leashw32.dll in afskfw library.
155 * NPLogonNotify must provide the user password in all
156 calls to KFW_AFS_get_cred(). It cannot count on a
157 credential cache being preserved between calls. Permits
158 tokens to be acquired for all cells listed in the
159 TheseCells registry value for a domain.
161 * Improve the trace logging from NPLogonNotify().
163 * Avoid a race when writing the cm_scache_t mountPointString
164 when acquiring mount point or symlink target data via
165 cm_GetData(). The race could result in bogus target
168 * Permit the use of des-cbc-md5 and des-cbc-md4 enctypes
169 as DES keys in asetkey.exe.
171 Since 1.5.9903 (1.6pre3)
172 * Fix automatic addition of Freelance cell mount points
175 * Avoid recursive offline volume checks which could cause
176 a crash due to stack exhaustion.
178 Since 1.5.9902 (1.6pre2)
179 * rx calls issued on a busy call channel are now specially
180 handled and logged. All calls are automatically retried.
182 * The vos support for multi-homed servers was incorrectly
183 implemented. The changes have been reverted.
185 * Avoid wasting cm_buf_t objects and associated data buffers
186 for obtaining mountpoint and symlink target strings. Instead
187 read them directly into the cm_scache_t mountPointTarget
190 * Optimize the evaluation of mountpoint and symlink targets
191 by skipping the FetchStatus and a full round-trip if
192 the cache manager does not have valid status information
195 * Fix the abstraction of cm_FreelanceAddMount() by passing
196 in volume names that do not include the trailing dot.
199 [there was no 1.6pre1 for Windows]
201 * vos commands now manipulate servers by UUID and can
202 recognize multi-homed servers.
204 * afs_config will not longer set the Tray Icon State
205 in the registry if the checkbox is not present in
206 the dialog. [RT 128591]
208 * Heimdal's roken utility library has been added
209 as \Program Files\Common\afsroken.dll
211 * When probing servers to determine if they are up or down
212 no longer issue an RXAFS_GetTime RPC to servers that indicated
213 that they do not understand the RXAFS_GetCapabilities RPC.
214 Since they responded it is known that they are up.
216 * AFS Explorer Shell Extension now works from folder
217 backgrounds. Overlays for mount points and symlinks
218 are present in the dll, but are not registered at present
221 * Do not use RankServerInterval registry value as the value for
222 PerformanceTuningInterval.
224 * CellServDB updated to 13 Dec 2010 release from grand.central.org
226 * Add "fs chmod" command and display current mode as part
227 of "fs examine" output.
229 * When the data version of a mountpoint or symlink changes,
230 the target string in the cm_scache_t object must be cleared.
231 Otherwise, the new target will not be queried.
233 * "fs checkservers" now includes vldb servers in the output
234 and only lists multi-homed servers once. A multi-homed
235 server that has at least one up interface is no longer
236 considered to be down.
238 * When asynchronously storing dirty data buffers to the
239 file server ensure that (a) the cm_scache_t object and
240 the cm_buf_t object are for the same File ID so that
241 locking and signalling work properly; and (b) if the
242 FID no longer exists on the file server, do not panic,
243 just discard the buffer.
245 * When processing VNOVOL, VMOVED and VOFFLINE errors perform
246 server comparisons by UUID or address and not simply by
247 cm_server_t pointer. Otherwise, server failover may not
250 * Do not preserve status information for cm_scache_t objects
251 when the issuing server is multi-homed.
253 * Giving up all callbacks when shutting down or suspending
254 the machine is now significantly faster due to the use
255 of an rx_multi implementation. (This functionality is
256 still off by default and must be activated by a registry
259 * Race conditions were possible when updating the state
260 of the cm_volume_t flags and when moving the volumes
261 within the least recently used list.
263 * Ensure that the lanahelper library does not perform a
264 NCBRESET of each lan adapter when enumerating the
265 current network bindings. Correcting this permits OpenAFS
266 to work on Windows 7 when the network adapter settings
269 * Fix creation of mount points and symlinks as \\AFS\xxxx
273 * Reimplement FetchData and StoreData operations to make
274 use of rx_Writev and rx_Readv which is faster than than
275 multiple calls to rx_Write and rx_Read. Throughput
276 improvements of up to 20% are possible.
278 * Many improvements to the rx rpc stack:
282 - Fast recovery is not triggered by packet resend timeouts
284 - Socket errors are now detected and reported
286 - avoid repetitive memory allocations
288 - avoid calling gettimeofday() for every packet
290 - reduced dependency on the call->lock
292 - When a call receive is completed, ack all of the packets
293 immediately. Do not wait for the first response packet
296 - Do not hold the call->lock during rx_Write* and rx_Read*
297 operations when the call fields in use can only be accessed
298 by the application thread.
300 - Reduce the default window size from 128 back to 32 because
301 the current overhead from walking packet queues in the rx
302 listener thread significantly hampers the ability to
303 process incoming packets from the network.
305 - Only backoff the peer timeout once for a series of naks.
307 - Release builds are lean and mean. All debugging and
308 packet tracking code is disabled.
310 * Version number is logged to the Windows Application Event
311 Log as part of the start pending message.
313 * MSI installers now backup and restore configuration
314 data during upgrades.
316 * Rx pmtu discovery has been disabled by default. Use
317 HKLM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
318 RxPMTUDiscovery "DWORD"
319 value to re-activate.
321 * cm_BkgDaemon event evaluation has been optimized to avoid
322 unnecessary overhead.
324 * If a volume is known to be readonly, then locally fail any
325 change request without sending it to the file server. This
326 avoids triggering the file server's abort threshold.
328 * Do not leak cm_volume_t objects from the volume LRU queue.
330 * Ensure that cm_NameI errors are acted upon promptly.
331 There are many cases in the SMB server where an error from cm_NameI()
332 was either ignored or not acted upon until several other operations
333 are performed that could result in the same error being repeated.
334 This is a mistake which did not have negative side effects until
335 additional checks for callback status were added recently.
337 At present, if a CM_ERROR_ACCESS error is returned and ignored,
338 subsequent attempts to operate on the same cm_scache_t will result
339 in additional queries to the file server that will also end in an
340 abort response. This can trigger the file server to delay responses
343 * Fix fs mount point and symlink create operations which were
344 failing due to improper directory separator termination.
346 * RXAFS_GetVolumeStatus can return VNOVOL, VMOVED, etc. In
347 order to process them and update volume state a fid must
348 be passed to cm_Analyze(). Use the volume root fid.
350 * Negative caching for volume lookups that fail with either
351 VL_NOENT or VL_BADNAME. The expiration time is five minutes.
352 This prevents volume lookup storms.
354 * Provide each volume lookup its own two hour lifetime instead
355 of reseting all volume location information every two hours.
356 This prevents the first smb path evaluation after the volume
357 location data has been discarded from potentially causing an
361 * When the SYSTEM account is in use, the SMB connection is
362 not always established as S-1-5-18. Sometimes it uses an
363 anonymous, S-1-5-7, connection. In both cases the SMB
364 authentication name is the nul string. To better identify
365 when an SMB connection is local system, the smb_username_t
366 is now stored as a SID instead of a user name. When a SID
367 is used, a new flag bit is set SMB_USERNAMEFLAG_SID.
369 In smb_SetToken if AFS_PIOCTL_AFSLOGON is set and the user
370 connection is not SYSTEM, the RPC SID is checked. If RPC
371 SID is SYSTEM, that supercedes the SMB authentication.
373 * Replace STATUS_IO_TIMEOUT with RPC_NT_SERVER_TOO_BUSY for
376 * Replace STATUS_PATH_NOT_FOUND with RPC_NT_SERVER_UNAVAILABLE
377 for CM_ERROR_ALLOFFLINE and CM_ERROR_ALLBUSY.
379 * Optimize behavior of buf_CleanAsyncLocked(). Avoid searching
380 for the cm_scache_t object by FID if it is already known.
382 * If the readonly file attribute is set (stored as a unix mode)
383 then a CreateFile operation should fail if the file is opened
384 for DELETE in combination with any other privilege.
386 * If the directory buffer contents are garbage we can crash
387 the service. Add some simple validation checks to ensure
388 that cm_dirEntry_t objects have the correct flag value and
389 that the name strings are not too long.
391 * If the bulkStat errorCode indicates that a particular object
392 is inaccessible due to a VIO error, we must update the server
393 status appropriately in order to permit failover.
395 * An RX_MSGSIZE error is returned by the new PMTU detection
396 code. It is critical that such an error result in a retry of
397 the operation that failed. Otherwise, the PMTU detection can't
398 work and the server will be marked down.
399 Secondly, it is important that such errors not leak to the
400 application layer. Map them to CM_ERROR_RETRY in all cases.
402 * Translate RX_RESTARTING to the same cache manager error code
405 * When logging server volume instance errors to the windows
406 application event log, be sure to log the cell as well.
407 Translating from server ip address is non-trivial. Make it
408 easier for administrators triaging issues to plug the volume
409 and cell info into vos commands.
411 * Since unix mode bits are represented in octal in most cases
412 make sure we log them that way.
414 * The cm_enforceTrailingDot() function failed if there was a
415 trailing dot followed by white space.
417 * Rx: Do not send RX_MSGSIZE if the PMTU size did not change.
420 * A crash could occur if a single letter share name was
421 dynamically evaluated as matching a cell name. This was
422 due to a failure to treat the comparison of two nul
423 strings as identical.
425 * Docbook validation is performed by xmllint during builds.
427 * Volume package bugs in the file server can result in VOFFLINE
428 being returned to the client instead of VNOVOL or VMOVED. As
429 a result the Unix CM treats VOFFLINE the same as VMOVED and VNOVOL.
430 The Windows client has not. As a result, bugs in the file server
431 can cause the Windows client to lose if the volume has in fact
432 been moved to another server.
434 As part of this change, the volume location list is updated prior
435 to the volume status being applied to the server from which the
438 * If the pages to be flushed are from a readonly or backup volume
439 they can't be dirty. In this case there is no need to stabilize
440 the pages before they are locked and recycled.
441 Stablilization is performed on the cm_scache_t object so do not
442 stabilize and unstablize for each cm_buf_t object.
444 * Over the years the processing of the Freelance callbacks have
445 added functionality that behaves much more like FetchStatus checks
446 to a file server. If the data version of the object has changed,
447 get the new data. Given that is the case, we can remove much of
448 the original refresh logic that is rather race prone. Say goodbye
449 to cm_fakeGettingCallback and cm_fakeDirCallback.
451 * When processing a pioctl path with either smb_ParseIoctlPath or
452 smb_ParseIoctlParent, cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK|GETSTATUS)
453 must be called on the cm_scache_t object to ensure that it is up
454 to date before we permit cm_Lookup or other operations to be performed
455 on it. Add the cm_SyncOp() call to smb_ParseIoctlPath and
456 smb_ParseIoctlParent to ensure it is done for all pioctl operations.
458 * The 32-bit tools installer was displaying client configuration wizard
459 pages. Since, the 32-bit tools share configuration with the AFS
460 service, the 32-bit tools installer should not be attempting to change
461 client configuration. Remove said wizard pages from the 32-bit tools
464 * cm_data.rootSCachep is a global pointer to the cm_scache_t that
465 represents the root.afs volume root directory. Throughout the
466 code this cm_scache_t was being used without ensuring that a
467 callback to the volume is in fact valid.
469 Under most circumstances this would not be a problem. However,
470 it is possible for a request to fail due to the lack of a callback
471 at a critical moment. Add a new function cm_rootScachep() that
472 attempts to ensure that a callback is present (if possible) prior
473 to use the cm_scache_t object.
475 * When the contents of the Freelance root directory changes the fake
476 directory buffers are updated and a fakeDirVersion is incremented.
477 The dataVersion of the cm_scache_t object is supposed to be updated
478 on the next access by performing a fake get callback request.
479 Unfortunately, this did not always occur because of a race. If another
480 Freelance object is updated first, the root directory object would
481 never successfully get a fake callback.
483 This patchset ensures that the generation of the fake directory
484 buffer content and the callback are obtained under the same set
485 of locks thereby removing the race.
487 * Enforce in afsd_service that tokens can be set for alternate
488 userids only if the SetToken pioctl is called from the
489 Local System account.
492 * Revise SMB QuerySecurityInfo Response for MS10-020
494 MS10-020 (http://support.microsoft.com/kb/980232) has caused
495 many problems for implementors of SMB 1.0 servers and applications
496 that call GetFileSecurity() without checking the return code to
497 determine if the call succeeded. The gist of the vulnerability
498 was that the SMB redirector would pass any buffer it received
499 to the application regardless of whether or not it was valid.
500 MS10-020 protects the applications by strictly validating the
501 SMB response data structure and the data in the security descriptor
504 The problem for SMB 1.0 server implementors is that there have
505 been at least three different protocol descriptions for
506 NT_TRANSACT_QUERY_SECURITY_DESC published over the last decade
507 and all of them are incomplete. Therefore, just about no one but
508 Microsoft has an SMB 1.0 server implementation that produces the
509 exact out that they are expecting to validate.
511 The end result is that in an attempt to protect applications from
512 crashing due to invalid input being passed in directly caused
513 dozens of applications to crash by not returning any security
514 descriptor data at all. Even when the applications didn't crash
515 they might not have been able to save their data. Cisco WAAS
516 and NetApp DataOnTap systems were most adversely affected and
517 they have had CIFS protocol licenses for many many years.
519 To fix OpenAFS here is what needed to be done:
521 1. Instead of returning a security descriptor that gives ownership
522 to the NUL SID, give it to the Everyone SID and set the flag
523 that states that everyone has full access.
525 2. Validate the input parameters. In particular, check to ensure
526 that the SMB file descriptor is valid and the file has not
529 3. Enforce the maximum output data and parameter counts.
531 4. Handle buffer overflow and buffertoosmall conditions
532 in the manner that Microsoft expects them to be handled.
533 In particular, note that the parameter data which is returned
534 in the SMB Data Region is not counted in the Data Count.
535 Even if MaxData is 0, we can still return parameters values
536 as long as MaxParm is large enough.
538 * Prevent use of AFSCache file contents if mapped to
541 * The Windows version of "fs newcell" did not accept any parameters
542 and behaved quite differently from the Unix version. Instead of
543 permitting new cell information to be added, the Windows version
544 simply forced the existing cell information to be reacquired.
546 This update adds a new pioctl, VIOCNEWCELL2, to support the
547 implementation of a Unix-style "fs newcell". The functionality
548 added here differs from the Unix version in the following ways:
550 1. "fs newcell" with no arguments is still accepted
551 in order to maintain compatibility with prior Windows
554 2. "fs newcell -cell <cell> -dns" instructs the cache manager
555 to add the new cell but obtain the vldb server info from
558 3. "fs newcell -cell <cell> ... -registry" instructs the cache
559 manager to add the new cell and also save the cell configuration
560 data in the registry for use the next time the service restarts.
562 4. The -vlport and -fsport options are accepted although the
563 -fsport value is currently unsupported by the cache manager.
565 * New registry value "FreelanceImportCellServDB" instructs Freelance
566 to create a mount point for every cell name listed within the
569 * Path MTU discovery for Rx is activated.
571 * Rx socket input buffer is converted to a circular buffer.
573 * Fix usage of cm_FreeServerList(). Do not set the server list
574 pointer to NULL after calling cm_FreeServerList(). Doing so
575 can result in a memory leak.
577 * Only enable Rx NAT pings on a single anonymous connection at a
580 * Fix cm_IoctlSkipQueryOptions() buffer management. Prevents a
581 potential read beyond end of memory buffer.
583 * Reduce requested privileges when reading registry CellServDB
584 to the minimum required.
586 * Add support for RPC Pipe Service NetWkstaGetInfo levels
587 101 and 102 which are called on Windows 7 and 2008-R2.
589 * Prevent integer overflow during quota percent used calculation
590 in Explorer Shell Extension (RT 126846)
592 * Generate a meaningful error if "fs listacls" or "fs setacls"
593 are executed on the Freelance root.afs volume.
595 * RXAFS_InlineBulkStat errors must be processed via cm_Analyze.
596 RXAFS_InlineBulkStatus does not return errors such as EACCES,
597 VNOVOL, VNOVNODE, VOFFLINE, VBUSY, VIO, VMOVED, etc. as an RPC return
598 code. Instead they are returned in the status info errorCode field
601 Traditionally, the error associated with the first FID in the query
602 list has been returned to the caller of cm_TryBulkStatRPC().
603 However, the error has never been processed through cm_Analyze()
604 which means that the per-vnode processing for VNOVNODE and the volume
605 global processing for VMOVED, VNOVOL, etc. has never been performed.
606 As a result, failover to other .readonly volume instances cannot occur,
607 volume moves will not be handled, and files that have been deleted
610 This patchset makes the following changes:
612 1. If an inline bulk operation has been performed and the inline
613 errorCode is a volume global error, then that error replaces
614 the RPC return code within the cm_Analyze() processing for
615 the RPC. This will affect whether or not a retry operation
618 2. The variable 'inlinebulk' is reset to 0 at the top of the
619 cm_Analyze() loop in case failover from an inlinebulk capable
620 file to an inlinebulk incapable file server takes place.
622 3. The FID that is passed into cm_Analyze() is not a real fid.
623 Instead it consists of the cell and volume but vnode = 0.
624 This ensures that the error (if any) is not applied to the
627 4. If an inline bulk operation was performed, prior to performing
628 the cm_MergeStatus() operation a vnode a check is made to
629 determine if an error was returned for that vnode. If so,
630 cm_Analyze() is called with no connection, a fake cm_req_t,
631 the fid, and the error. This permits cm_Analyze() processing
632 to be performed on the file.
634 * Show configuration pages for all types of MSI installations
636 The OpenAFS MSI installer wizard used to not show any configuration
637 pages for "Typical" and "Complete" installations. Setting the
638 workstation cell and logon options during installation required
639 selecting the "Custom" option. Many users choose the "Typical" option
640 during installation, and thus would never see the configuration pages.
641 Therefore, for these users, the workstation cell was being set to the
644 This patch makes the workstation cell and logon option configuration
645 pages visible to all types of installations (except silent
646 installations which show no UI).
648 * cm_LookupInternal creates Freelance mount points and symlinks
649 when queries cannot be found in the Freelance root.afs directory.
650 If the search name is a full cell name for which vldb information
651 can be obtained, then a mount point is added. If the search name
652 is a left-most substring or the full cell name with a dot appended
653 to it, then a symlink was created. This approach created a very
654 poluted Freelance name space.
656 This patchset makes the following changes:
658 1. Do not create symlinks with a dot appended to the cellname
660 2. Do not create symlinks where the left-most substring is not
661 a full dot separated component of the cellname.
663 3. Permit lookups to succeed when we would have created a
664 symlink in the past without creating the symlink.
666 * BPlus tree lookups are much faster than searching through
667 the native directory format on Windows because the case sensitive
668 hash tables cannot be used successfully. Permit BPlus trees
669 to be used except when called with cm_BPlusDirFoo as the action
670 function because cm_BPlusDirFoo is used to build the BPlus trees
671 from the native directory format.
673 * Symlinks are ending up in the Freelance root.afs directory that
674 end with a dot. Make sure it cannot happen.
676 * cm_FreelanceAddMount and cm_FreelanceAddSymlink is supposed to
677 return the allocated FID of the entry that was added. However,
678 cm_NameI is called to perform the lookup without forcing an update
679 of the Freelance fake directory. As a result the entry may not be
682 Force an update prior to calling cm_NameI() by using
683 cm_clearLocalMountPointChange() and cm_reInitLocalMountPoints()
686 * The Freelance fake root directory buffers were not zero-filled.
687 This results in random behavior that can cause the service to
688 terminate unexpectedly.
690 * The validation check for the response from the GetVolumeStatus
691 pioctl is incorrect. The response is not simply a VolumeStatus
692 structure but also several C strings appended to it.
694 * When flushing a file, we need to commit the file length changes
695 as well as the dirty buffers. Call cm_FSync instead of buf_CleanVnode
696 which is called by cm_FSync.
698 * Prevent rx_rpc_stats global lock from being a bottleneck in the
703 * Avoid a race when updating cell vldb server lists
704 that can result in a crash.
706 * Avoid a deadlock when managing CM_SCACHESYNC_STOREDATA
707 state operations for directory objects.
709 * Add new Windows Application Event log messages for
710 VBUSY, VRESTARTING, ALL_BUSY, ALL_OFFLINE, and ALL_DOWN.
711 Include message throttling to prevent the same message
712 from being logged repeatedly within a five second window.
714 * Reduce lock contention by waiting for cm_buf_t I/O
715 operations to complete before permitting cm_SetupStoreBIOD
716 to analyze a buffer for inclusion in a BIOD.
718 * Split the cm_buf_t flags field to separate the flags
719 that are protected by the cm_buf_t mutex from those
720 protected by the buf_globalLock. This eliminates the need
721 to hold both locks everytime the flags field is accessed.
722 Both locks were not held in the past resulting in race
723 conditions that could result in deadlocks.
725 * Add "vos setaddrs" command.
727 * Rx library lock contention avoidance between rx_NewCall and
730 * Rx library races due to inconsistent use of rx_connection
731 conn_data_lock to protect the flags field.
733 * Rx library inconsistent use of RX_CALL_TQ_WAIT which could
736 * Rx library must signal transmit queue waiters when flushing.
737 Otherwise, deadlocks can occur.
739 * In cm_UpdateVolumeLocation, avoid searching for a ".readonly"
740 volume on a numeric volume name.
742 * File buffer allocations whose offsets are beyond server EOF
743 should be locally allocated and zero filled. The file server
744 should not be issued a FetchData rpc which is guaranteed to
747 * Enable integrated logon to work with Windows 7/2008 when
748 user logons are performed with a non-Domain Kerberos principal.
750 * Add Protection Error messages to aklog output.
754 * Prevent the Explorer Shell extension from crashing if
755 symlink creation failed. (126406)
757 * A Rx level NAT ping has been implemented.
758 Add NatPingInterval registry value to
759 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
760 to permit Nat Ping to be enabled. The default value is 0 seconds.
762 * When a re-initialization is taking place, be sure to reset
763 cm_noLocalMountPoints to 0 in case someone deletes the "Freelance"
764 registry key out from underneath the service.
766 * Add krb5 error message translation to aklog, afscreds,
767 afslogon.dll, the network identity manager afs provider
770 * Mode bits aren't directly exposed by the Win32 API. We were leaving
771 them to default to 0777 when creating new files and directories.
772 This version introduces two configuration parameters;
773 'UnixModeFileDefault' and 'UnixModeDirDefault' which are DWORD
774 registry entries that are used to set the initial mode bits.
775 If the values are set to 0, then the behavior is identical to what we
778 * Minidump files are now produced with a timestamp appended
781 * An SMB request debugging monitor has been added. When activated
782 the monitor will automatically turn on trace logging if any SMB
783 request has required longer than 60 seconds to complete and will
784 then create a minidump every 60 seconds thereafter until the
789 * Restore use of DNS AFSDB and SRV records by kaserver clients.
793 * Avoid a potential Freelance deadlock during initial execution
794 of afsd_service.exe if the old ini file data has to be
797 * Three rx library corrections. (1) Idle data connection
798 processing could timeout if the send window filled and
799 took longer than the idle data timeout period for the
800 transmit window to re-open. (2) The transmit queue
801 could be emptied prematurely. A required check for the
802 queue being in use was forgotten. (3) The function that
803 is supposed to implement a wait for the transmit queue
804 to cease being busy failed to wait.
807 * Restore use of DNS AFSDB and SRV records which were
808 unintentionally disabled in 1.5.69
811 * Add a context menu to the NetIdMgr AFS Provider
814 * Prevent an empty directory Btree from being created and
815 marked as valid if cm_BPlusDirBuildTree fails.
817 * cm_BPlusEnumAlloc should not fail if the enumeration
818 contains zero entries.
820 * In cm_BPlusXXX functions, return ENOMEM if malloc() fails.
822 * Do not leak the cm_scache_t dirlock if cm_BPlusDirBuildTree
825 * fs examine should report owner and group ids as signed values.
827 * Protect buffers in smb_WriteData from simultaneous writes.
829 * Query network interface MTU values from the registry because
830 the IP Helper GetAdapterAddresses() fails to report a manually
833 * Improve performance by not dropping and reacquiring the rx_call
834 lock in rx_WriteProc* and rx_ReadProc*.
836 * When configuring the rx library, the network interface MTU
837 must be applied to both the maximum send size as well as the
838 maximum receive size as is done when rx_SetMaxMTU is called
841 There was no 1.5.67 release
844 * Convert afsauthent.dll to use the xdr_alloc and xdr_free
845 routines from afsrpc.dll in place of malloc and free.
846 This permits the two libraries to be built with different
847 C runtime libraries without crashing.
849 * Set the DOS Readonly attribute on a file/directory
850 whenever the unix mode combined with the mask 0200
851 is true. Previously there was a discrepency between
852 the mask used for testing for readonly behavior and
853 that used for setting the attribute.
855 * Disable AFSVolSync based .readonly "whole-volume callback"
856 support because the all file servers prior to 1.5.67
857 (and perhaps 1.4.12) do not properly assign a value to
858 the AFSVolSync structure in bulk status RPC responses.
860 * Improve the error output from aklog to output the value
861 from krb5 error_message() if the afs_com_err output
862 indicates an unknown value.
864 * Hold a lock on the cm_scache_t object when the smb
865 server is assigning a new clientModTime value.
867 * Permit custom build version numbers to be set
868 without modifying the src/config/NTMakefile.<platform>
871 * Add support for binary code signing with cross-signed
872 certificates and arbitrary versions of signtool.exe.
874 * Convert VBUSY and VRESTARTING to CM_ERROR_ALLBUSY and
875 do not permit them to be exposed to the smb redirector.
877 * Convert STATUS_TIMEOUT responses to STATUS_IO_TIMEOUT
878 to avoid confusion within the smb redirector.
880 * Fix the byte order assigned to port numbers associated
881 with AFSDB record lookups. They must be network byte
882 order not host byte order.
884 * Add dynamic server ranking based on RPC round trip
888 * Official support for Windows 7 and Server 2008 R2
890 * Improved Rx RTT computation following the lessons
893 * Prevent a file server bug (FetchData returning an
894 invalid length instead of zero) from causing an
895 "unexpected network error" when writing to files.
897 * Promote DNS SRV records as superior to DNS AFSDB
898 records. Support arbitrary port numbers for vldb
901 * Add "fs listacl -cmd" support.
903 * Add AFSVolSync based .readonly "whole-volume callback"
904 support. With this functionality, multiple objects from
905 a .readonly volume can have their status validated by
906 issuing a single RXAFS_FetchStatus RPC.
908 * Remove drive mapping functionality and service start/stop
911 * Remove drive mapping functionality from afs_config.exe.
913 * Adjust SMB error return codes to avoid returning
914 STATUS_TIMEOUT which results in the SMB redirector
917 * Network Identity Manager OpenAFS Provider now provides its
918 own "AFS lock" notification icon to report the status of
919 "have tokens, have no tokens, service not started, service
920 started but inaccessible". Hovering over the icon lists the
921 cells for which tokens exist (if any) and the OpenAFS version
922 number. Double-clicking executes the Network Identity Manager
925 * When merging the current status for an AFS object, ensure
926 that the volume status for that object is set to 'vl_online'.
928 * Permit cm_Analyze to request a retry even when the cm_req_t is
929 flagged CM_REQ_NORETRY if the RPC failed due to RX_CALL_DEAD.
930 The retry will force the use of a new connection.
932 * Add digital signatures to resource only dlls as required by
935 * Prevent pioctl calls from retrying indefinitely when a sharing
936 violation error occurs.
938 * Add {HKLM,HKCU}\SOFTWARE\OpenAFS\Client DWORD "ShowMountTab"
939 to restore access to drive mapping functionality in afscreds.exe
942 * Permit Rx minimum peer timeout, maximum receive window size and
943 maximum send window size to be configured via the registry.
945 * The MSI installer now sets the ARPINSTALLLOCATION as required by
948 * DNS lookups set the port numbers in network byte order within
951 * Replace use of the 16-bit compatibility API WinExec with
952 ShellExecuteEx in afs_config.exe and afscreds.exe. WinExec
953 is incompatible with Windows 7 / Server 2008 R2 UAC
957 * Short circuit background volume checks if a shutdown
958 or suspend request is received.
960 * Avoid contacting the file server in order to allocate
961 an empty buffer beyond the current length of the file.
963 * When a request for streams on a directory or mount
964 point object is received, do not offer a default stream.
965 This was affecting the saving of roaming profiles.
968 * Fix Windows 2000 compatibility. Do not call rand_s()
969 even if it is compatible with the C run time library
972 * Fix a data consistency error between the output of
973 NetWkstaGetInfo and NetServerGetInfo RPCs, specify the Lan
974 workstation group name "AFS", and report server name as
975 "AFS" instead of "\\AFS" when the caller asks for "\\AFS".
977 * Enable executables to be run from \\AFS on Windows 7.
978 Return "Name not found" instead of "File not found" when
979 a directory or file name cannot be found.
981 * Prevent cache manager from marking file server "down" when
982 the data returned in response to either RXAFS_FetchData64
983 or RXAFS_StoreData64 is invalid.
985 * Add pioctl data validation to the AFS Explorer Shell extension.
988 * After 1.5.62 afslogon.dll would no longer crash but it
989 also would not load "domain" specific configurations.
992 * An uninitialized variable was present in the symlink
993 recursion detection loop.
995 * Reverse addition of unique per file GUID during
996 Create/Open responses. Inclusion of the GUID prevents
997 Cygwin applications (and others) from accessing //afs.
999 * Treat "filename::$DATA" as a synonym for "filename".
1000 This addresses the recent report of roaming profiles
1001 failing to save back to the file server.
1003 * When given a choice between an error from rx_Write()
1004 or RXAFS_EndStoreData(), use the rx_Write() error.
1005 This ensures that over quota and access denied errors
1008 * If an error occurs during RXAFS_StoreData operations,
1009 apply that error to all of the buffers in the BIOD.
1010 This ensures that failures will not be retried if they
1011 will only fail again.
1013 * Do not perform a DNS lookup for a cellname if the
1014 cellname prefix is _._AFS_IOCTL_._. This will avoid
1015 DNS lookups when the query is for
1016 _._AFS_IOCTL_._.F7E5F580200909010061TTTTNT7TT.{10E39A49-4531-4496-A08E-842D4C440D20}.
1018 * Fix Freelance root.afs volume object callback processing.
1019 Object synchronization and status merging bugs.
1021 * Fix Explorer Shell "invalid parameter" bug introduced
1024 * Log RXKAD errors to the trace log.
1026 * Prevent simultaneous pioctl calls from multiple processes
1027 from stepping on each other's data.
1029 * Prevent simultaneous pioctl calls from crashing the
1032 * Add data validation to pioctl output processing in
1033 ktc_GetToken and ktc_ListTokens.
1035 * Add data validation to pioctl output processing in
1036 fs.exe and symlink.exe.
1039 * IMPORTANT: New support for DCE RPC Services: SRVSVC and WKSSVC.
1040 No longer will browsing \\AFS produce truncated share names.
1042 * Registry specified server preferences did not result in
1043 the CM_SERVERFLAG_PREF_SET flag being applied to the
1046 * Avoid unnecessary DNS lookups of share names as cells
1047 when it is known that the name cannot be a cell name.
1048 Any name that does not contain a dot is skipped.
1050 * When processing Dfs Referral requests, do not return an
1051 error if the server is down or busy; if the volume is
1052 offline; the cell vldb cannot be reached; etc.
1053 These paths are still \\AFS paths and so the client
1054 should not be sent to search elsewhere for resolution.
1056 * CRITICAL: Prevent the cm_Daemon thread from terminating when
1057 the machine enters the suspend state. IF the cm_Daemon thread
1058 dies, there is nothing to execute down server checks.
1060 * Prevent use of smb_StartedLock before initialization which
1061 can result in a panic in the lock package.
1063 * Improve error checking in afslogon.dll in order to prevent
1064 NULL pointer dereferences within GetLogonDomainOptions()
1065 if the Lsa operations fail.
1067 * Add support to the build system to automatically update
1068 a Microsoft Symbol Store.
1070 * Do not access the cm_conn_t after executing the cm_Analyze
1071 loops. Doing so can result in an access to memory that has
1074 * CRITICAL: If a StoreData request offset is not aligned on the
1075 buffer module blockSize, (offset % blockSize) bytes of dirty
1076 data will not be written to the file server.
1078 * CRITICAL: If a StoreData64 request is sent to a file server
1079 that does not support large files (>2GB) and the client is
1080 unaware that the server has no large file support, the first
1081 chunksize worth of data will not be written to the file server.
1082 This bug was introduced in 1.5.3. It affects all IBM AFS file
1083 servers and all OpenAFS file servers older than 1.4.0.
1086 * If a file server becomes inaccessible while the cache manager has
1087 dirty buffers to write, the afsd_service buf_IncrSync thread can
1088 attempt to use 100% of the cpu.
1090 * Fix the locking that protects the cm_cell_t lists
1091 (all, name, id, free).
1093 * Remove cm_cell_t objects from the name and id hash tables before
1094 placing them on the free list.
1096 * Fix "fs newcell" which was broken in 1.5.60.
1098 * Do not attempt to synchronize dirty buffers if the associated
1099 volume is known to be unavailable.
1101 * Log to the Windows Event Log dirty buffers that cannot be written
1102 during the shutdown process.
1104 * Modify behavior of a Freelance mountpoint target that does not
1105 specify a cell. Instead of assuming the target volume is in the
1106 Freelance.Local cell, use the workstation "Cell" specified in the
1107 registry. A mountpoint target of "#root.cell." will now mean the
1108 root.cell volume in the workstation cell for the current session.
1109 If the workstation cell changes from "athena.mit.edu" to
1110 "andrew.cmu.edu", the referenced volume will also change without
1111 requiring that the mount point targets be altered.
1113 * Avoid false infinite loop errors when validating the cm_cell_t
1114 free list during startup.
1116 * Fix two more locations where xdr_free() should be used instead of
1117 free() to permit a checked built of afsd_service.exe to work with
1118 release builds of afsrpc.dll.
1120 * Add cm_FindServerByUuid(). Re-implement RXAFS_InitCallBackState3()
1121 to permit the server Uuid to be used to lookup the server object
1122 and from that determine the cell. This permits callbacks that are
1123 received from alternate addresses to be processed with a known server
1124 object. Previously a request from an unknown server would clear all
1125 callbacks from all cells.
1127 * Fix a bug that prevented optimal performance when using a non-zero
1128 value for 'daemonCheckVolCBInterval'. As a reminder, when
1129 "daemonCheckVolCBInterval" is set to a non-zero value, all .readonly
1130 volume callbacks are automatically renewed 90 minutes before their
1133 * Fix automatic ranking of vldb servers whose values are obtained from
1134 the CellServDB file.
1136 * Add logging for RX CALL DEAD errors which are generated if the remote
1137 host is unreachable.
1139 * Add failover for RX CALL TIMEOUT errors when the volume is readonly
1140 or the call is to a vldb server.
1142 * Add registry based cell search functionality to NetIdMgr,
1143 afs_config.exe, and klog.exe.
1145 * afsconf_GetCellInfo() has been modified to perform gethostbyname()
1146 lookups on the host names in the CellServDB instead of using the
1147 specified IP addresses. This provides aklog, pts, vos, etc. the same
1148 CellServDB behavior that the Windows Cache Manager uses.
1150 * Completely a behind the scenes change. The snprintf() routines have
1151 been completely rewritten.
1153 * When updating the stat cache entry callback of a .readonly object
1154 from the volume group object, update the file server reference to
1155 ensure it matches the most update to date callback.
1157 * Add proper support for processing callbacks from multi-homed file
1158 servers. Instead of comparing servers by cm_server_t pointer,
1159 compare them by UUID when the UUID is known.
1161 * During a shutdown short circuit the offline volume check daemon
1164 * Properly track link counts during cm_Link and cm_Unlink operations.
1166 * Return the error code of RXAFS_FetchData / RXAFS_StoreData in
1167 preference to an error code reported by rx_EndCall.
1169 * Add "PerFileAccessCheck" registry value to permit testing against
1170 experimental file servers that include per-file acl support. This
1171 value is intentionally undocumented. It is not to be used by
1172 production environment deployments.
1174 * Log a warning to trace log if a volume group is marked as a DFS
1175 File Set. The Windows OpenAFS client does not include DFS support.
1177 * Fix a bug introduced in 1.5.60 that prevents the afs netidmgr
1178 provider from obtaining tokens when referrals are in play.
1180 * Add "fs chown" and "fs chgrp" commands to permit the owner and
1181 group of objects stored in AFS to be set from Windows.
1183 * Avoid performing background daemon operations when the machine is
1184 going into suspend mode.
1186 * Perform offline volume checks in most recently used order.
1188 * Prevent crash when a data version for a cache object goes backwards.
1190 * fs setquota, fs setcachesize, vos setfields, and vos create now
1191 accept human readable orders of magnitude. (K, M, G)
1193 * fs listquota fixed to permit large quota sizes to be displayed.
1195 * rx packet resend and data packets sent counts were incorrect.
1197 * Multi-thread safe library versions are now being generated and used.
1198 mtafsubik.lib, mtafsutil.lib, mtafsvldb.lib, mtafsvol.lib.
1200 * Microsoft SMB Redirector (mrxsmb.sys) support for ExtendedSessTimeout
1201 values are now available on XP through Windows 7. Add functionality
1202 to autodetect if such support is present on the machine. If so,
1203 configure it if necessary and dynamically adjust the AFS Rx timeout
1206 * Correct another race condition in the Rx library that could result
1207 in an unexpected panic while freeing the Rx call iovq.
1210 * A fix to the pioctl library to support drive substitution
1211 to UNC paths. (SUBST <d:> <\\afs\cell\path>).
1213 * On April 9th Microsoft released a Hot Fix for Windows Server 2003 SP2
1214 that corrects a deadlock in the smb redirector and also adds new
1215 functionality that permits the AFS SMB server to be given a longer
1216 timeout than is normally the case. New functionality has been added
1217 to configure these additional LanmanWorkstation\Parameter values.
1218 (This functionality has been backported to XP SP3 and is scheduled
1219 to be released on June 5th.)
1221 * The BackConnectionHostNames registry value configuration was broken
1222 when dynamic re-establishment of Netbios Name registrations was added.
1225 * Hidden vos commands are revealed.
1227 * If the "DisableLoopbackCheck" registry value is set, do not unset
1228 it during the same service session.
1230 * Reorganize code that prevents multiple Store operations to the same
1233 * Modify IsPathInAfs test in Explorer Shell Extension and fs.exe to
1234 permit broken symlinks to be treated as being in AFS.
1236 * Fix vos commands that output 64-bit integer values.
1238 * Cygwin Import Libraries are provided in the SDK for all OpenAFS DLLs.
1239 Permits building cygwin applications against OpenAFS libraries.
1241 * OpenAFS Release Notes, Administrator Guide and User Guide now installed
1242 as Windows HTML Help (.CHM) files.
1244 * NSIS installer does a much better job of cleaning up files left
1245 over from previous installs.
1247 * Fix RT#124787, a race condition between "fs flush <dir>", "fs flushvolume",
1248 or "fs flushall" and on-going directory operations that can result in
1249 afsd_service.exe crashing.
1251 * Add support for DNS SRV records in place of AFSDB records.Â
1252 _afs3-vlserver._udp.<cellname>. Priority field is used.Â
1253 Weight is currently ignored.
1255 * Add a method of specifying Client CellServDB information within the
1256 registry that can be used to either override the CellServDB file or
1257 force the use of DNS lookups for a given cell.
1259 The registry schema is as follows:
1261 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\
1262 "LinkedCell" REG_SZ "[cellname]"
1263 "Description" REG_SZ "[comment]"
1264 "ForceDNS" DWORD {0,1}
1266 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\[servername]\
1267 "HostName" REG_SZ "[hostname]" Default: [servername]
1268 "IPv4Address" REG_SZ "[address]" Used only if gethostbyname() fails.
1269 "IPv6Address" REG_SZ "[address]" <future>
1270 "Comment" REG_SZ "[comment]"
1271 "Rank" DWORD "0..65535" Default: 0
1272 "Clone" DWORD "{0,1}" <future: server only>
1273 "vlserver" DWORD "7003" <future>
1274 "ptserver" DWORD ... <future>
1276 ForceDNS is implied non-zero if there are no [servername]
1277 keys under the [cellname] key. Otherwise, ForceDNS is zero.
1278 If [servername] keys are specified and none of them evaluate
1279 to a valid server configuration, the return code is success.
1280 This prevents failover to the CellServDB file or DNS.
1282 Only one of "HostName" or "IPv4Address" is required. "HostName"
1283 is optional if [servername] is the host name.
1285 * Extend registry based CellServDB functionality to
1286 afsconf_GetCellInfo() interface used by aklog and many
1287 other command line utilities.
1289 * libafsconf.dll moved from OpenAFS\Client\Program to OpenAFS\Common
1290 as it is now used by both client and server components.
1293 * PriorityClass of afsd_service.exe process raised to
1294 "High" to match the priority of the system services
1295 that are waiting for afsd_service.exe to respond.
1296 Configurable using [TransarcAFSDaemon\Parameters]
1297 DWORD "PriorityClass" value. (use values from Win32
1300 * Change error reporting of CM_ERROR_BADSHARENAME to
1301 use the correct error (NTSTATUS_BAD_NETWORK_PATH).
1302 The error we were sending would cause the smb
1303 redirector to drop the connection.
1305 * Change the error reporting of CM_ERROR_ALLOFFLINE
1306 and CM_ERROR_ALLDOWN to return NTSTATUS_PATH_NOT_FOUND.
1307 Previous we would return Bad Network Path which
1308 would cause the smb redirector to cancel all file
1309 handles on the same "share".
1311 * A debug version of afsd_service.exe can now be used
1312 with a release version of afsrpc.dll (or vice versa).
1313 afsrpc.dll now implements and exports xdr_free().
1315 * Removed all unprotected OutputDebugString() calls as
1316 those calls can block for up to 10 seconds if there
1317 is contention for the globally shared memory or if
1318 the debugger fails to read data in a timely manner.
1320 * Visual Studio 2008, WDK 6.1 and DDK 6.1 are now
1321 supported for builds. The production releases still
1325 * Discovered that XP/2003 supports a variant on recvmsg
1326 called WSARecvMsg. The input parameters are mostly
1327 compatible with the Posix recvmsg. This will save
1328 multiple memcpy() calls per received packet.
1330 * Discovered that Vista/2008 supports a variant on
1331 sendmsg called WSASendMsg. The input parameters
1332 are mostly compatible with the Posix sendmsg. This
1333 will save multiple memcpy() calls per sent packet.
1335 * Exporting rx_ReadProc32, rx_WriteProc32 from afsrpc.dll
1336 and utilizing rx_ReadProc32 from within the cache manager
1337 as part of the FetchData operations. rx_ReadProc32 and
1338 rx_WriteProc32 are optimized versions of rx_ReadProc and
1339 rx_WriteProc for use when reading 32-bit integer values.
1341 * The Network Identity Manager AFS provider now reports
1342 progress events. These provide a better user experience
1343 with Network Identity Manager 2.0.
1345 * [RT 124293] A race condition exists which permits the
1346 scp field of the an smb_fid_t object to become invalid
1347 after a request on the smb_fid_t has begun. Remove the
1348 race by protecting all access to the scp field with the
1349 smb_fid_t mx mutex and obtaining a local reference on
1350 the cm_scache_t object for the length of the request.
1352 * [RT 124276] If the vldb is out of sync with the contents
1353 of the file servers, afsd_service will retry too many
1354 times when a file server reports a volume as not being
1355 present. Now if the list reported by the vldb is the
1356 same as the previously seen list, then the retry is
1359 * [RT 124276] Read-only volume failover was broken in
1360 1.5.53 whenever accessing a volume results in VNOVOL or
1363 * [RT 124276] Prior to 1.3.70 the volume server reference
1364 list was not reference counted and would be prematurely
1365 freed while in use. When reference counting was added
1366 in 1.3.70 a bug was introduced that could result in
1367 service reference list corruption.
1369 * Add Windows Application Event Log warning messages for
1370 "Client SMB MPX value too large" and
1371 "Client SMB Buffer Size too small".
1373 * Renaming of files across directory boundaries would result
1374 in an invalid handle error when attempting to access the
1375 files after the move.
1377 * Fix the handling of Tran2 Set Path Info RPCs. Do not fail
1378 when a smb file descriptor cannot be found. The whole point
1379 of using a Path Info function is because an smb file
1380 descriptor wasn't allocated.
1382 * More edge cases in which dynamic addition of Freelance
1383 root.afs entries would get the wrong FID or where the
1384 root.afs directory would not be refreshed.
1386 * Buffer overflow could occur if the workstation cell name was
1387 longer than 64 characters. Crashes could occur in
1388 afscreds.exe, afslogon.dll, and afsd_service.exe.
1390 * VNOSERVICE and VOFFLINE errors were leaking and were exposed
1393 * Improve consistency of the bad data version value used for
1396 * Log file server uuid values as part of the cm_server object
1397 when available. Dump the cm_server object list in response
1400 * Optimize the performance of resetting access control lists
1401 when tokens are set or removed.
1403 * Fix vos.exe so that it can work with very large volume
1406 * Remove symlink recursion tests and increase max symlink count
1410 * Undo the "UAC manifest fix" applied to afs_config.exe.
1412 * Fixed some additional race conditions in the Rx library
1414 * Ensure that Freelance allocation of vnodes follow
1415 the AFS convention of odd vnodes are directories
1416 and everything else is an even vnode.
1418 * Add Freelance logic to mount point and symlink
1419 evaluation functions.
1421 * Enhance smb_ParseASCIIBlock() so that it can handle
1422 all of the STRING formats defined by the CIFS Technical
1425 * Validate the output of smb_ParseASCIIBlock() in all
1426 callers. Return CM_ERROR_BADSMB if the STRING field
1427 cannot be parsed. CM_ERROR_BADSMB will cause the
1428 contents of the packet to be logged.
1430 * If multiple SMB Raw Write operations were taking
1431 place at the same time, there could be data corruption
1432 because unique event objects were not generated for
1433 each Netbios receive operation.
1437 * Fix a race in rxi_WriteProc which could result
1438 in a freed or in use packet being passed to
1439 rxi_PrepareSendPacket if the call is currently
1440 in a transmit wait state. Passing a freed packet
1441 will result in a panic. Passing an in use packet
1442 might result in a call failure or perhaps data
1445 * Kerberos referrals broke the NIM AFS provider
1446 logic for automated configuration of the workstation
1447 cell with a new identity. Now if a referrals
1448 response is returned, assume the cell belongs to the
1451 * The interface to buf_CleanAsync() and
1452 buf_CleanAsyncLocked() has been changed to permit
1453 errors to be propagated to the caller.
1455 * During an enumeration of a directory, if the btree
1456 has not yet been generated, return an error.
1458 * When mapping CM_ERROR values to NTSTATUS values
1459 do not treat 0 as an error.
1461 * Avoid a cm_scache_t object reference count underflow
1462 when renaming via the SMB interface.
1464 * smb_IsMask() can be called with a non-8.3 input buffer.
1465 Separate smb_IsMask() into smb_Is8Dot3Mask() for 8.3
1466 input buffers and smb_IsMask() for non-8.3 input buffers.
1468 * Add cm_buf_t error state to the "fs memdump" output.
1470 * pioctl drive mapping checks need to take into account
1471 Global Drive Mappings and SUBST assignments.
1473 * Rename operations should not be permitted if the
1474 destination name collides with an existing name that
1475 differs only by case UNLESS the collision is the object
1476 that is being renamed.
1478 * Increase the maximum number of background daemon threads
1479 to 64. Leave the default at 4.
1481 * Set the FILE_UNICODE_ON_DISK flag when reporting the
1482 file system characteristics.
1484 * Report the maximum component name length as 255 even
1485 for Unicode as part of the file system characteristics.
1487 * Add RX debugging information to "fs memdump" output
1490 * Refcount the cm_server_t objects returned by
1493 * Fix the UAC manifest applied to afs_config.exe
1495 * rxdebug now indicates if any RX_CONN flag is set
1496 not just some of them
1498 * Change the mapping of CM_ERROR_WOULDBLOCK from
1499 NTSTATUS_LOCK_NOT_GRANTED to NTSTATUS_CANT_WAIT
1503 * Includes #GCO Public CellServDB 07 Nov 2008
1505 * Prevent separate cell entries from being created for
1506 "foobar.org" and "foobar.org." which can happen if
1507 an AFSDB record is found instead of a CellServDB
1510 * Fix a panic when checking the SMB virtual circuit list.
1511 If a VC is destroyed while the list is being checked
1512 it is possible for the next object to become invalid.
1514 * Increased logging to afsd_init.log of netbios name
1517 * Fix a DNLC panic when a UTF-16 string that cannot be
1518 converted to a utf-8 string is provided as a filename.
1520 * In the NSIS installer, delete old afslogon.dll/afscpcc.exe
1521 from %windir%\system32 and fix the WinLogon\Notify\AfsLogon
1522 location to refer directly to the new location in
1523 %programfiles%\openafs\client\program
1525 * Be more aggressive about responding to Netbios
1526 NRC_BRIDGE/NRC_NOWILD errors. Rebind the Netbios name to
1527 the LANA within the listener thread without performing a
1530 * Use the WNet Enumeration of all connected drives as a filter
1531 in the pioctl GetIoctlHandle operation. This should remove
1532 the last situations in which there was a delay when opening
1533 an Explorer Shell properties menu with the AFS Shell
1536 * Fix the parsing of CellServDB in Windows to handle linked
1537 cells as is done on Unix.
1539 * Add support for linked cells to the cache manager,
1540 netidmgr afs provider, and aklog. When two cells are
1541 linked, a volume lookup that fails in one cell is looked
1542 up in the other. This can be used for construction of
1543 test cells that override specific volumes; cell renaming;
1546 * Increase the length of cell and realm names that can
1547 be entered in the NIM AFS provider.
1549 * Actively prevent a 32-bit afsd_service.exe from
1550 executing in the WOW64 environment.
1552 * Protect against invalid IP address values in CellServDB
1557 * Fix an Rx panic caused by an attempt to double free
1558 an rx_packet when an error occurs while writing the
1561 * Fix an exception cause by a race condition when
1562 processing multiple file change notifications registered
1563 by the Explorer Shell or other applications.
1565 * Modify the Rx debug interface to use var_args and
1566 permit Rx debugging to be used on 64-bit Windows
1567 which previously had it disabled.
1571 * Automated lock hierarchy enforcement added to the
1572 osi base lock package. This permits detection of lock
1573 violations that might but often does not result in
1574 deadlocks. Enabled by default in the checked builds.
1575 May be activated in release builds via the registry
1576 value "LockOrderValidation" in the TransarcAFSDaemon
1579 When a violation is detection an assertion is thrown
1580 generating a dump file and a message indicating which
1581 locks were obtained out of order is logged to the
1582 Windows Application Event Log.
1584 * Attempt to ensure that a dump file will be generated
1585 whenever the service panics.
1587 * Removed unused lock cm_bufGetMutex
1589 * Many out of order lock acquisitions were detected
1590 using the lock order verifier and were corrected.
1592 * A race condition within the function path cm_GetSCache() ->
1593 cm_GetNewSCache() -> cm_RecycleSCache() permitted a
1594 cm_scache_t object to be issued simultaneously to two
1595 threads. This would eventually result in a panic due
1596 to the resulting undercount.
1598 * Fix interpretation of the empty string as the ioctl path
1599 to mean the current directory. "fs lsm", "symlink list",
1600 etc. now return a "not a ..." error instead of "does not
1603 * Add SMB_STRF_SRCNULTERM flag to smb_ParseStringXXX functions.
1604 to indicate that the input string is a nul terminated string.
1605 Assign ti when input strings are nul terminated instead of
1606 length counted. (from 1.5.50)
1608 * The CIFS protocol specification for handling NT_TRANSACT_CREATE
1609 does not match the observed behavor. The 'nameLength' is specified
1610 in bytes, not in characters. Fix the implementation to match
1611 the observed behavior. (1.5.50)
1613 * The cm_HaveAccessRights() attempt at deadlock avoidance by
1614 calling lock_TryRead() on the parent directory cm_scache_t rw-lock
1615 does not avoid the deadlock. Avoid the deadlock by enforcing the
1616 lock order of the lowest vnode first. Then remove the infinite
1617 loop avoidance in cm_SyncOp() that was returning an unwarranted
1618 access denied error.
1620 * File Server lock synchronization was not propertly enforced in
1621 smb_ReceiveNTCreateX() and smb_ReceiveNTTranCreate().
1623 * Another change to the short name generator. Plus and equal are
1624 not valid characters in a short name according to the
1625 RtlIsNameLegalDOS8Dot3() in kernel mode. They have been replaced
1626 with left and right brace in order to minimize the impact.
1628 * Fix a race condition that can result in a heap overwrite error if
1629 the number of known file servers or vldb servers increases during
1630 the preparation phase of a cm_CheckServers() operation.
1632 * If the SMB Authentication would fail, a LSA allocated memory
1633 buffer would be leaked.
1635 * Rewrite the cm_cell_t allocation code to permit cell vldb server
1636 lookups to fail without leaking cm_cell_t objects.
1638 * Add "fs setcrypt auth" to permit the use of rxkad authentication
1639 only mode. Not documented.
1641 * Initialize Rx mutex and lock objects once and only once as part of
1642 the DLL load process.
1644 * When reporting the Volume Name "AFS" do not include the terminating
1645 NUL in the reported length.
1647 * Refresh Volume Location information upon receipt of a VNOVOL error
1648 from the file server and avoid a race condition that prevented the
1649 updated volume location information from being destroyed immediately
1650 after it was acquired.
1652 * In response to VNOVOL and VMOVED, remove the volume id from the server's
1653 volume list so that the volume's up and down status is not affected
1654 when the machine is shutdown.
1656 * Many fixes to the Rx stack related to packet allocation and deallocation;
1657 Rx debugging; Rx packet tracking across queues; and Rx packet loss caused
1658 by overwriting the rx_call currentPacket pointer when it was already
1659 referring to a packet.
1661 * Rx Jumbograms are now off by default. To re-enable, add "RxJumbo" = "1"
1662 in the TransarcAFSDeamon/Parameters key.
1664 * "fs flush", "fs flushvolume", and "fs flushall" now destroy locally
1665 constructed directory B+ trees for flushed directory objects.
1667 * When cached pages for directory objects have been locally modified, do
1668 not permit the locally modified pages to be mixed with pages subsequently
1669 obtained from the file server. Doing so can result in directory
1672 * If Ubix errors are received in response to VL_xxx operations record
1673 the error on the server instance and permit the operation request to
1674 be retried in case another server is able to answer the query. We have
1675 seen at least one instance of a cell in which a VL server returned
1676 UNOQUORUM (possible database truncation) when the other servers were
1677 responding successfully.
1680 * Optimize the assignment of write errors to dirty
1681 buffers by applying the first error to all buffers
1682 belonging to the same object instead of trying each
1685 * Add resource version info to xstat_fs_test and xstat_cm_test
1687 * Fix a delete vs delete[] usage problem in the WiX custom
1690 * When creating a mount point, validate the volume name to
1691 make sure it exists.
1693 * When processing a mount point, pay attention to VL_BADNAME
1696 * Remove a race condition from cm_volume_t recycling that could
1697 have resulted in two threads making use of the same object at
1700 * Fix a reference count leak on smb_vc_t objects introduced in
1701 1.5.50. Permit the objects to be freed once again.
1703 * Fix the output of "fs memdump" for smb_vc_t objects that was
1706 * Add routines for debugging smb refcount errors.
1708 * When shutting down the cache manager, do not permit AFS
1709 callback requests to be processed. Doing so can result in
1710 invalid memory accesses.
1712 * Do not recycle a deleted cm_scache_t object unless the
1713 refCount is 0. Doing so can result in a deadlock and
1714 invalid memory access.
1716 * Minimize the number of RPCs sent to the file server for
1717 directory bulk status requests. Only request status for
1718 objects that we are actually reporting on and only if we
1719 do not already have status for the object and then make
1720 sure that we return as much as we can at once via the SMB
1721 interface but not so much that it exceeds the RDR timeout
1723 * Enable bos restricted operations
1725 * Fix the creation of submounts used by the afscreds and afs_config
1728 * Avoid repeated retries when attempting to obtain access rights.
1729 If the access rights were obtained successfully but do not
1730 satisfy the request, fail the request.
1732 * Fix error code reported when attempting delete a file on a
1733 readonly volume or one that is marked with the readonly DOS
1736 * Fix a short name truncation bug.
1738 * Fix a heap corruption error when reading the CellServDB file
1741 * Fix directory listings when the directory has "lookup" permission
1742 only and the directory has never been previously accessed.
1744 * Fix several tracking issues for cm_scache_t rw locks.
1746 * Ensure that all buffers used to store cell names are the
1749 * Restore Windows 2000 support (broken in 1.5.50).
1750 LOCALE_INVARIANT was introduced in XP.
1752 * Add RxUdpBufSize registry value. Default is 256K.
1754 * Avoid multiple deadlocks between cm_freelanceLock and cm_scacheLock
1756 * Do not include trailing NULs when returning file names in
1757 a directory search response.
1759 * Prevent RX from crashing the afsd_service by attempting
1760 to allocate to many buffers at once when memory is fragmented.
1762 * Avoid a deadlock when freeing cm_server_t objects
1764 * Optimize search cell file. Do not search to end of file
1765 if the cell has already been found.
1767 * Remove explorer shell extension delay on local disks.
1769 * Automatically pre-allocate 64 Rx packets per thread
1771 * Add smb lock requests and stat cache lock allocations to
1772 the output from fs memdump.
1775 * Fix Large File Support (broken in 1.5.50)
1777 * Fix "fs setacl" (broken in 1.5.50)
1779 * Fix pioctl path evaluation of drive letter
1780 mapped paths. (broken in 1.5.50)
1782 * Fix installation of VC8 Runtime library in the EXE
1783 Retail installer. (broken in 1.5.50)
1785 * Avoid a potential deadlock during Volume Location
1786 updates if all the file servers are unreachable at
1787 the time the VL update succeeds.
1789 * Ensure that rx connection object is references
1792 * Fix propagation of over quota errors during cm_FSync()
1795 Since 1.5.39 [1.5.50 released 16 July 2008]
1796 * There were no releases between 1.5.39 and 1.5.50.
1798 * FIRST RELEASE WITH UNICODE SUPPORT
1800 * Pioctl interfaces to the cache manager refactored
1801 to provide layering between the SMB specific code
1802 and the general purpose ioctl operation.
1804 * Garbage collect dead SMB virtual circuits as soon
1805 as they are no longer being referenced. This avoids
1806 problems with outstanding locks not being dropped
1807 when the virtual circuit becomes invalid.
1809 * Remove the IBM Administration Reference documentation
1810 and replace it with the OpenAFS Command Reference
1813 * Avoid calling rx_SetDeadTime and rx_SetHardDeadTime
1814 functions each time a connection is about to be used.
1815 Do not hold a lock on the rx connection object while
1816 it is being selected. This avoids a race between
1817 threads attempting to set the timeout values and
1818 removes a bottleneck that was hampering performance.
1820 * Ensure that the smb directory attribute is set
1821 for all directory objects.
1823 * Add vs2008 support to the NSIS installer scripts
1825 * Replace the VC Runtime EXE installer with the MSI
1826 installer in the NSIS installer scripts
1828 * Properly delete the VC7.1 C Runtime libraries.
1830 Since 1.5.36 [1.5.39 released 23 June 2008]
1831 * There were no 1.5.37 or 1.5.38 releases for Windows
1833 * Use Visual Studio version of vsprintf() instead of
1834 home grown version that resulted in linking conflicts
1836 * Prevent buserver, ptserver and fileserver from crashing
1837 due to non-portable strftime() format strings.
1839 * Prevent server crashes caused by rx_InitMorePackets()
1840 being called prior to allocation of all of the required
1841 synchronization objects.
1843 * Fixed a memory leak in the internal B+ tree directory.
1844 For objects with non-8.3 names, the 'longname' memory
1845 allocation would be leaked during object deletion or
1848 * Fixed a memory leak in the background daemon cm_CheckServer()
1849 routine. The memory allocated to store the server list
1852 * Obtain a missing lock around a call to cm_RemoveSCacheFromHashTable().
1854 * Correct an abstraction layer violation. cm_scache_t objects
1855 should be marked deleted in cm_Unlink() and cm_RemoveDir() and not
1856 in smb_CloseFID(). Cleanup of deleted cm_scache_t objects should be
1857 performed in cm_ReleaseSCache() when the reference count hits zero.
1859 * Prototype cm_AdjustScacheLRU() and re-implement it using osi_QAddH().
1861 * Do not reference 'smb_logp' in cm_vnodeops.c. Use 'afsd_logp'
1864 * Ensure that scache object 'nextp' pointers are set to NULL
1865 when the objects are removed from the hash table linked list.
1867 * If a dirty buffer cannot be written to the file server because
1868 of an access denied error, mark the error in the cm_buf_t
1869 structure so that it can be dealt with instead of retrying
1873 * Update CellServDB file to GCO Public 23 Apr 2008
1875 * Fix a cm_buf_t reference count leak when attempts to write
1876 dirty buffers to the file server from within cm_IncrSyncer()
1879 * Prevent udebug from crashing.
1881 * Activate RX Idle Timeouts. If the file server is busy for more
1882 then 30 seconds, attempt to failover to another server without
1883 marking the busy server down.
1885 * Another VNOVNODE issue fixed. When writing a dirty buffer
1886 to the file server, if VNOVNODE is received, mark all buffers
1887 as invalid without further attempts to contact the file server.
1889 * Improved performance on high latency links.
1892 * Fix a bug preventing the re-initialization of the Freelance
1893 root.cell contents when a change is made.
1895 * Properly return CM_ERROR_NOSUCHFILE when removing a symlink
1896 or mountpoint that does not exist.
1898 * aklog, afscreds, integrated logon, and the NIM AFS credential
1899 provider will all attempt to first acquire an afs service ticket
1900 using the service principal: afs/<cell>@<CLIENT_REALM>. This
1901 permits the proper detection of the cell realm when multiple
1902 realms are local realms for the cell.
1904 * Fix fs commands that re-use file identifiers instead of forcing
1905 a path evaluation each time. This was broken in 1.5.33 when
1906 hashes were added to the cm_fid_t structure.
1908 * Fix symlink and mountpoint removal and then recreation when
1909 applied to the fake Freelance root.afs volume.
1911 * Fix SMB change notifications for the Freelance root.afs volume
1912 when creating or removing symlinks and mount points.
1914 * Force DNS AFSDB cell name lookups to lower case if the DNS
1915 response is not all lower case.
1917 * 32-bit installers are now built with Visual Studio 2005
1919 * All vos.exe commands now support the "-noresolve" option which
1920 forces the display of IP addresses instead of DNS hostnames.
1922 * No longer hold the cm_volume_t mutex across VL_xxx RPCs. This
1923 permits operations that do not require up to date volume
1924 location data to continue while waiting for the RPC to complete.
1925 Multiple threads waiting for a volume location update on the
1926 same volume set no longer trigger multiple calls VL_xxx RPCs.
1928 * Volume server references could be marked offline and never
1929 reset to the not_busy state.
1931 * Further optimizations to cm_Analyze() retry logic for
1932 CM_ERROR_OFFLINE and CM_ERROR_ALLBUSY error handling.
1935 * Optimize the DNLC by applying Interlocked incrementing for
1936 statistics gathering and enabling greater use of read
1937 locks instead of write locks
1939 * Further optimize osi_Log operations when logging is disabled
1940 by removing the need for the function call overhead.
1942 * Add a new lock operation, Convert Read to Write, and apply
1943 it throughout the cache manager. This function provides a
1944 fast transition from a read lock to a write lock when the
1945 caller is the only reader.
1947 * Fix a timeout problem when opening files or request locks
1948 when the file in question is already write-locked by another
1951 * Fix a deadlock that can occur if "fs flushall" is executed
1952 while applications are writing to AFS.
1954 * The cache manager would re-read data from the file server
1955 after file truncation when obtaining buffers to write to the
1956 same file until the first StoreData RPC completes.
1958 * Fixes error handling in cm_NTCheckDelete()
1960 * Replaces cm_scache_t mutex with a read-write lock permitting
1961 additional parallel access.
1963 * Prevent an cm_scache_t reference undercount when evaluating
1964 symlinks containing @sys when none of the active sysnames
1965 match any of the available targets.
1967 * Prevent the leak of a cm_scache_t rwlock if associated symlink
1970 * Set RxMaxMTU registry value to 0 which is the equivalent of
1971 it not being set at all.
1973 * Reduce contention between BkgDaemon threads
1975 * Prevent CM_SCACHEFLAG_ASYNCSTORE from being reset on a write
1978 * Convert reference counts on cm_volume_t, cm_conn_t, cm_server_t
1979 objects to use the faster Interlocked operations which avoid
1980 the need to hold write locks.
1982 * Permit rxkad errors other than RXKADEXPIRED to be treated as a
1983 non-fatal error that forces a retry against another service
1986 * Fix a crash in smb_WriteData that would occur if an application
1987 opened the file as read-only and then sent a write request anyway.
1990 * Respond to VL_NOENT errors by removing cm_volume_t object and
1991 setting it to be the first object to be recycled. This is not
1992 only a more efficient use of resources but it also prevents
1993 repeated attempts to query the VLDB server for the volume from
1994 the CheckOfflineVolumes routine.
1996 * Change the defaults for RxEnablePeerStats and RxEnableProcessStats
1999 * A first cut at a cache manager statistics monitor that can be used
2000 to determine the necessary cache parameters to support the
2001 application working set.
2003 Off by default, the performance package can be activated by setting
2004 "daemonPerformanceTuningInterval" in the service Parameters key.
2005 As with the other daemon interval values the unit is in seconds.
2007 At service start and each succeeding interval the cache manager
2008 will write statistics to %TEMP%\afsd_performance.txt showing the
2009 relative usage of cm_scache_t, cm_volume_t and cm_buf_t objects.
2010 The FID statistics keep track of all FIDs seen by the cache manager
2011 during the service session whether or not they are backed by any
2012 live object in the cache.
2014 These statistics are not stored in the cache file.
2016 * Fix windows event logging broken in 1.5.33.
2018 * Eliminate multiple calls to syscfg_GetIFInfo() which is used to
2019 populate the IP address list for WhoAreYou/TellMeAboutYourself RPCs
2020 as well as when computing the server preference list.
2022 * When merging status after a StoreData operation prevent data
2023 buffers from being discarded that we would prefer to keep.
2024 (Bug introduced in 1.5.33.)
2026 * Modify installers the set RxMaxMTU to 0.
2028 * In the AFS Configuration Panel on the Advanced->Miscellaneous
2029 dialog, the Background Daemons and Service Threads fields were
2032 * Increase default number of Background Daemons to 4.
2034 * Remove cm_volume_t reference from the cm_scache_t object. Permit
2035 cm_volume_t objects to be recycled.
2037 * Windows Error Reporting has shown a number of previously unseen
2038 stack corruption errors in the Explorer Shell extension. The
2039 minidump appeared to indicate the problem was in ParseAcl().
2040 Reviewing the function it appears that numerous things could go
2041 wrong if invalid input was provided. Added error checking and
2042 data validation. Hopefully this will address the problem.
2044 * Windows Error Reporting has shown crashes in aklog as a result
2045 of krb5_cc_get_default() failures not being handled. They are
2049 * The Rx library used a 32-bit type for sockets which was
2050 truncating the socket value on 64-bit Windows. This
2051 was not an issue in general but became an issue when
2052 Rx "hot thread" support was enabled.
2054 * Enable "Rx hot thread" support which ensures that while
2055 a received message is being processed that a new thread
2056 is allocated to receive the next message to arrive.
2058 * Increased parallel access to smb_rctLock protected
2059 resources. Reduce the number of times that the
2060 lock is dropped and re-obtained.
2062 * Increased parallel access to cm_scache_t and cm_buf_t
2063 resources by replacing the previous global write lock
2064 protected reference counting mechanism with one built
2065 upon the use of InterlockedIncrement and InterlockedDecrement.
2067 * Implement a hash for cm_fid_t to reduce comparison costs
2068 and inline the cm_FidCmp() function.
2070 * Redefine the BUF_HASH and BUF_FILEHASH in terms of the cm_fid_t
2071 hash which has a better distribution
2073 * Modify cm_ConsiderPrefetch to evaluate the amount of data in
2074 the most recent read request instead of the next chunkSize.
2075 cm_chunkSize can be dozens or hundreds of buffers. As a result
2076 too much time is spent performing the evaluation.
2078 * Fix the usage of cm_scache_t bufCreateLock. The purpose of this
2079 lock is to prevent the creation of new buffers while a truncation
2080 is being performed. All references to bufCreateLock have been
2081 removed except in two places:
2082 i. a write-lock surrounding the function that calls buf_Truncate()
2083 ii. a read-lock within buf_GetNewLocked() that actually allocates
2086 * Modify the CM_CONFIG_DATA_MAGIC value to include a
2087 CM_CONFIG_DATA_VERSION number which value be used to force the
2088 replacement of the cache file contents when incompatible changes
2089 are made between releases.
2091 * CM_SCACHESYNC_ASYNCSTORE should not be ordered by
2092 cm_SyncOpCheckContinue
2094 * Avoid calls to multi_Rx if nconns == 0
2096 * Modify smb_WriteData to perform background writes based upon the
2097 crossing of 'smb_AsyncStoreSize' boundaries instead of cm_chunkSize
2098 boundaries. This will slow down writes from the SMB interface but
2099 will avoid the risk of the CIFS client disconnecting from the AFS
2100 client SMB server. The default value for 'smb_AsyncStoreSize'
2103 * SMB AsyncStore functionality is on by default but can be
2104 disabled using the registry value:
2106 HKLM\SOFTWARE\OpenAFS\Client DWORD "EnableSMBAsyncStore"
2108 smb_AsyncStoreSize can be adjusted with
2110 HKLM\SOFTWARE\OpenAFS\Client DWORD "SMBAsyncStoreSize"
2112 The value must be a multiple of the buffer block size
2113 and cannot be larger than the chunk size.
2115 * Do not leak a read lock on the bufCreateLock
2117 * Do not forget to call cm_SyncOpDone if cm_SyncOp succeeded
2120 * In MergeStatus, only remove clean buffers from the hash table.
2122 * Avoid the need to update the data version number on each buffer
2123 associated with a scache when MergeStatus is called after a
2124 StoreData by maintaining a range of valid data versions as part
2125 of the cm_scache_t object.
2127 * Profiling shows large numbers of blocked calls in cm_HoldSCache
2128 from cm_HaveAccess... We can safely avoid the call so do so.
2130 * Add "ForceAnonVLDB" service parameters registry value. Default
2131 is 0. Set to 1 to force the use of anonymous connections to
2132 the VLDB servers. This is useful when using krb5 tokens and the
2133 VLDB servers have not been updated.
2136 * Do not attempt to search for a cell name that matches one
2137 of the special share names "svrsvc", "wkssvc", "ipc$"
2139 * Issue permission denied warnings from afscreds if attempts
2140 to modify the startup shortcut or registry values fail.
2141 This happens under Vista UAC.
2143 * Do not return BAD_NETWORK_ERROR in response to attempts
2144 to create directories or files in the Freelance root.afs
2145 volume. Instead return ACCESS_DENIED.
2147 BAD_NETWORK_ERROR was being sent because the Freelance cell
2148 has no vldb servers. Short circuit the error by testing for
2149 freelance in CreateFile and MakeDir
2151 * Treat srvsvc, wkssvc and ipc$ as pseudo files. Permit them
2152 to be opened and closed without actually existing. For now
2153 we treat them as equivalent to ioctls but this should
2156 * Fix RAP processing to not reject names that exist in the
2157 root.afs volume for GetInfo queries.
2159 * In afscreds, use GetUserNameEx to obtain the username instead
2160 of searching the registry. Try UserPrincipalName first and
2161 fallback to SamCompatible.
2163 * Stop linking to rpcns4.dll which is no longer supported on
2164 Vista and 2008. TaAfsAdmSvr is going to have to be
2167 * When processing dfs referral requests, if the sharename is
2168 a partial match and therefore will not be added automatically
2169 to the share list, make sure that we return an error.
2171 * Avoid a crash when logging VNOVOL errors returned from
2172 cm_CheckOfflineVolume.
2174 * When trace logging is disabled, do not permit osi_LogSaveString
2175 to perform the string copying. By disabling the string
2176 copies, throughput of the service more than doubles.
2178 * when checking offline volumes, update the cell vlserver info
2179 if required and don't perform a check if there are no servers
2182 * Add volumes in the vl_unknown state to the list of objects to be
2183 checked as offline or retried when an allbusy error is received
2186 * Add support for configuring rx_ExtraPackets
2188 * All calls to buf_Get() must be protected by a read lock on the
2189 cm_scache_t bufCreateLock
2191 * When a MergeStatus results in the invalidation of buffers, do
2192 not remove buffers from the hash tables that have active references.
2194 * Upon receiving a VNOVNODE error from a file server, be sure to
2195 invalidate any file locks. This prevents on-going attempts to
2196 extend the locks and will cause subsequent attempts to access
2197 the file to receive a invalid handle error.
2199 * Prevent smb_ReceiveTran2SearchDir from adding terminating NULs
2200 to directory entry names.
2202 * No longer discard tokens when RXKADBADTICKET error is received.
2203 Instead return STATUS_NO_KERB_KEY error.
2206 * Add "cmdebug <host> -cellservdb" option and client service
2207 implementation permits querying the vldb server for each
2208 cell known to the cache manager. The output is in a form
2209 that permits it to be used as a CellServDB file.
2211 * Add Symlinks->Show option to the AFS Explorer Shell Context
2214 * Prevent buffer overrun in the cmd parser used by command
2215 line tools such as fs, vos, pts, etc.
2217 * The path ioctl operations have several issues:
2219 (1) the specified path for "fs examine, whereis, whichcell,
2220 flush" and so always has the follow symlinks and mount
2221 points semantics. This makes it impossible to determine
2222 what the FID of a symlink or mount point is.
2224 (2) "fs examine" out is not the result of a single pioctl
2225 operation but is actually the combined output of half
2226 a dozen operations. Path evaluation is an expensive
2227 operation. It would be faster if the caller could
2228 evaluate the FID first and then perform all of the
2229 rest of the operations by specifying the FID instead
2232 (3) fs output reports all objects as files. By adding a
2233 GetFileType pioctl more informative output can be provided
2234 that indicates what type of object the path evaluates to.
2236 (4) the Windows fs command includes a number of commands that
2237 do nothing but exist only because the Unix cache manager
2240 A new extendible data structure cm_ioctl_query_opts_t has been
2241 added which can be optionally specified with pioctls that do not
2242 already require input data. The first two fields of this structure
2243 are 'literal' and 'fid'. The literal field is used to indicate
2244 whether the last component of the path should be evaluated following
2245 symlinks and mount points.
2247 The fid field permits a fid to be specified.
2249 * A new GetFileType pioctl has been added. The type of objects are
2252 * A new "-literal" option is available for "fs examine, flush, whereis,
2255 * Unimplemented fs commands have been removed.
2257 * There are circumstances where a volume object is being accessed and
2258 the volume is marked indicating that the volume location information
2259 is out of date but where it is also pointless and perhaps dangerous
2260 to block waiting for the rpc to complete. One example is when
2261 processing the cmdebug requests. If we know that we are not going
2262 to use the volume object to contact a server, then we can now set
2263 the CM_GETVOL_FLAG_NO_RESET flag.
2265 * when moving up in the directory tree we search the recorded fid list
2266 to find a matching fid that we have already crossed. we must also
2267 reset the fid count based upon what we discover.
2269 * Cell names are published as share names. As a result they are searched
2270 for as part of the Dfs Referral evaluation. If share "foo" can not be
2271 reached, the CIFS client will fallback to searching for "fo" as well.
2273 Since the freelance client automatically adds a symlink for prefixes
2274 of the cell name, this results in both "foo" and "fo" being added to
2275 the freelance root.afs volume. This patch prevents that negative side
2278 * when tracking Kerberos credential cache names be sure to include
2279 the cache type prefix.
2281 * Microsoft has assigned OpenAFS a network type value
2283 * The algorithms that were used to produce a Netbios name were broken.
2284 The name that was produced when the hostname was too long was in fact
2285 longer than the maximum permitted netbios name.
2287 Also, the "NetbiosName" value was not used as a suffix when the loopback
2288 adapter was not installed. The hardcoded string "AFS" was used instead.
2290 * avoid another deadlock during server probes initiated by the ipaddr
2291 change daemon thread.
2293 * prevent the afs client service from crashing when fs newcell is
2296 * In the NSIS installer, move afslogon.dll, afscpcc.exe and afs_cpa.cpl
2297 to \Program Files\OpenAFS\Client\Program in order to get them out
2298 of the %windir% tree
2300 * EDQUOT == WSAEDQUOT. Define it in the right places so that in the
2301 end STATUS_OUT_OF_QUOTA can be returned to the application.
2303 * Modify the search order for determining the location of CellServDB
2304 and other client configuration files.
2306 1. AFSCONF environment variable
2308 3. NEW - use All Users\AppData\OpenAFS\Client only if CellServDB exists
2309 4. use Program Files\OpenAFS\Client
2311 * More Vista Power Management tweaks
2313 * Close a socket handle leak in the kauth routines
2315 * Don't hold a cm_volume_t mutex across RPCs
2317 * Convert downTimes to use clock time instead of relative times.
2318 This permits correct comparisons with expiration times.
2320 * Update the Wix installer to properly install the prerequisites
2321 for translate_et.exe
2323 * Do not translate pioctl paths as they are not converted by the
2326 * Implement multi_rx probes for checking whether servers are up
2327 or down. Turns probes into a constant time activity instead
2328 of one dependent upon the number of known servers.
2330 * Treat VNOVOL errors the same as VOFFLINE. Do not force an
2331 update of the volume location data and retry. VNOVOL means
2332 that the volume could not be attached. It doesn't mean that
2333 the volume is not on this server.
2335 * Modify the Explorer Shell AFS Context menu Symlink->Add to
2336 not attempt to validate the target path. There is no reliable
2339 * Improve performance of \\afs\<cell><mount-type><volume>\
2340 access by eliminating extraneous attempts to search the
2341 CellServDB file and DNS AFSDB records. Share names
2342 containing '%' or '#' are not valid cell names.
2344 * Return STATUS_RANGE_NOT_LOCKED if the application attempts
2345 to unlock a range that is not currently locked.
2347 * Enable the 32-bit tools to read the 64-bit registry keys
2348 when executing in the WOW64 environment.
2352 * Fix buffer refcount leak introduced in 1.5.29
2354 * Prevent the NIM AFS Provider from crashing when configured to use
2355 Kerberos v4 for token acquisition in situations where no Kerberos v4
2356 support is available. For example, 64-bit KFW or sites that delete
2357 the Kerberos v4 DLL.
2359 * In afsd_init.log, support logging cache sizes greater than 2GB.
2361 * The volume status output of "fs examine" was being determined based
2362 upon the wrong error value. Instead of using the pioctl() return
2363 code, the errno value must be used.
2365 * Prevent invalid pts auto-registration attempts by aklog and the
2366 NIM afs provider when Kerberos referrals are in use. (krb5 1.6+)
2368 * Add synchronization protection to all of the SMB Listener State
2369 variables. This prevents race conditions when the addition or
2370 loss of a network adapter takes place.
2372 * On Vista, there is a race condition between the restart of the
2373 afs client service and the network adapters. Ensure that the
2374 afs client service can handle switching between loopback and
2375 non-loopback modes of operation.
2377 * Fix FollowBackupPath to work with mount points that are not
2378 explicitly read/write.
2380 * Fix a deadlock that has been observed on Vista while resuming
2381 after at least four hours of sleep.
2383 * Add new HKLM registry configuration options that can be used
2384 to provide configuration hints to the NIM AFS Provider when
2385 creating new identities. See release notes for details on
2386 HKLM\SOFTWARE\OpenAFS\Client\Realms\.
2389 * Speed up write buffer operations by holding a global read-lock
2390 instead of a write-lock when updating the buffer data version
2391 number. The data version is protect by the associated stat
2394 * Speed up read buffer operations by removing out of date buffers
2395 from the buffer hash tables.
2397 * Protect against a null pointer dereference in afslogon.dll
2398 if QueryAdHomePathFromSid() fails.
2400 * Remove a race condition while accessing private freelance data
2403 * Update the buffer data versions when Freelance directory data
2406 * Complete the implementation of smb_ReceiveTran2GetDFSReferral().
2409 * Return EAS_NOT_SUPPORTED in response to EA read/store/query
2410 instead of returning zero length.
2412 * Return ERROR_RETRY when a server sends VBUSY.
2414 * Return PATH_NOT_COVERED instead of BADSHARENAME for DfsLinks when
2415 a volstat plug-in is in use.
2417 * Enhance the volstat interface to support Dfs path mappings.
2419 * Fix "symlink list" to do the right thing for Dfs links
2421 * Fix callback expiration that was broken in 1.5.27.
2423 * Fix a crash at shutdown caused by the lack of synchronization
2424 within the daemon threads (cm_Daemon, cm_BkgDaemon, cm_IPAddrDaemon)
2426 * Hold the cm_server_t mutex while walking or modifying the
2427 cm_server_vols_t list in order to avoid a race condition.
2429 * Ensure that the cm_server_t refCount is safely modified.
2431 * The pthread library implementations of pthread_cond_wait
2432 and pthread_cond_timedwait were failing to return with
2433 the mutex held when error conditions occurred. This could
2434 result in race conditions wherever they are used. One case
2435 in which problems were seen was the Rx event_handler().
2437 * Prevent cm_GetBuffer() from locking all buffers associated with
2438 a stat cache object before deciding if the current buffer is up
2441 * Fix cm_BkgPrefetch to actually ensure that there is a cm_buf_t
2442 object for each offset that we will attempt to reserve.
2444 * Reduce the default number of background daemon threads used to
2445 prefetch or store to one in order to prevent two or more threads
2446 from stepping on each other.
2448 * Reduce the default chunk size to 256KB in order to prevent read
2449 timeouts on low bandwidth connections.
2451 * Fix a race condition that could result in threads becoming stranded
2452 when synchronizing multiple requests on the same stat cache object.
2455 * The GiveUpAllCallBacks functionality that was added in 1.5.21
2456 has been disabled. It has been disabled because it turns out
2457 that the RPC is improperly implemented in all file servers
2458 previous to 1.4.6. The RPC is executed without the proper
2459 locks being held resulting in data corruption and eventual file
2460 server failure. The functionality will be re-enabled in a
2463 * a couple of .readonly Volume Callback optimizations were
2465 - Apply most recent volume callback to all stat cache objects
2466 in the volume. This avoids premature callback expirations
2467 and unnecessary FetchStatus calls.
2468 - Add option to permit attempts to renew .readonly Volume
2469 Callbacks on a periodic basis to avoid invalidation of stat
2470 cache data. [Requires daemonCheckVolCBInterval be set to
2471 non-zero number of seconds. 1800 (30 minutes) is suggested.]
2473 * Adds support for 64-bit data version values. Previous releases
2474 only supported 32-bit values. Larger values sent by the file server
2475 would result in rollover.
2477 * aklog and asetkey no longer generate an exception if Kerberos for
2478 Windows is not installed.
2480 * If the LSA authentication functions report a temporary out of memory
2481 condition, fail the current authentication and do not cause the AFS
2482 client service to terminate unexpectedly.
2484 * When the AFS client service is shutting down, prevent the daemon check
2485 thread from executing additional checks which might attempt to access
2486 discarded locks or unmapped memory.
2488 * Correct a double free condition in the NIM AFS credentials provider.
2490 * Add "FollowBackupPath" service parameter registry value.
2491 This provides equivalent functionality to the UNIX afsd -backuptree
2492 option. When set to a non-zero value, normal mount points originating
2493 in a .backup volume will prefer a .backup volume.
2496 * Fix NSIS uninstall from the .MSI installer.
2498 * Add Stack Frame logging for AMD64
2500 * Add support for EWOULDBLOCK to cm_Analyze. If the file
2501 server returns EWOULDBLOCK, retry the request every two
2502 seconds for up to the RDRtimeout.
2504 * Reorganize the order of the includes to ensure that EWOULDBLOCK
2505 is not assigned the same value as EIO.
2507 * Update CellServDB to GCO Public 25 Oct 2007
2509 * Add additional validation and error handling code after
2510 each call to getSlot(). If an invalid slot is returned,
2511 return NONODE. If the invalid slot is returned when
2512 extracting a data node, invalidate the tree.
2514 * Modify compareKeys() to always perform a case-insensitive
2515 comparison and only perform a case sensistive comparison
2516 if the case-insensitive one matches. This ensures the
2517 ordering is consistently reported.
2519 * Add lock assertions to ensure that all calls are being
2520 performed with the correct locks being held. There have been
2521 some crash reports that provide stack data that does not appear
2522 to be possible unless there is a race. However, there are no
2523 obvious locations where the race is taking place and the test
2524 suite indicates that all of the correct locks are being held.
2525 We shall see what happens in the field.
2527 * For consistency replace all calls to findKey in which the range i
2528 s (1,numentries) with calls to getSlot().
2530 * Optimize the depth search loop by testing the slot value in the
2531 for statement instead of forcing the loop to be broken later.
2533 * Reorganize the locking for cm_BeginDirOp and cm_EndDirOp.
2534 There are a number of locations where locks are obtained, dropped,
2535 and reobtained. This reorganization attempts to accomplish several
2538 (1) be optimistic for the most common case so it will be fast
2540 (2) add consistency checks after each location where locks are
2541 dropped and re-obtained. If we lose a race in cm_BeginDirOp
2542 and the bplus tree is out of date, retry until we get to a
2543 consistent state that we can use.
2545 (3) Ensure that all operations take place with the correct locks.
2547 * One of the issues that has become a serious problem since the
2548 addition of the local directory updates is that although cm_SyncOp
2549 synchronizes operations, it does not preserve the order of requests.
2550 This has always been a problem in that it has been possible for a
2551 request to fail to complete due to its worker thread's bad luck.
2552 When a request takes longer than the Windows SMB Redirector's timeout,
2553 the SMB Redirector tears down the SMB virtual circuit.
2555 When using the local directory updates it is really important that
2556 the directory update operations complete in the order that they were
2557 sent to the file server. If they don't, then the local directory state
2558 and the file server state will not match and the local directory state
2559 must be discarded which in turn forces a new read of the entire
2560 directory contents over the network.
2562 This patch adds a new cm_scache_waiter_t object that is used to store
2563 the current thread, buffer, and syncop flags within a waiters queue
2564 on each cm_scache_t object. If a thread is forced to sleep in
2565 cm_SyncOp, upon waking it will check to see if there are any other
2566 threads waiting that are attempting to perform a similar task ahead
2567 of it in the queue. If yes, the thread goes back to sleep. If not,
2568 it goes ahead and enters the cm_SyncOp conflict resolution block.
2570 This patch has the additional side effect of reducing the number of
2571 competing threads that must obtain the cm_scache_t mutex and process
2572 the cm_SyncOp conflict resolution block. As a result, the overall CPU
2573 utilization of the service and the clock time associated with processing
2574 requests will be reduced.
2576 * assert that the cm_scache_t mutex is held when calling
2577 buf_ForceDataVersion().
2579 * rename findNode to leafNode in bplus_Lookup
2581 * replace all OutputDebugString calls with osi_LogX calls
2583 * modify bestMatch to special case the return values for leaf nodes.
2584 If an entry is above or below the values available in the leaf node
2585 return BTLOWER or BTUPPER instead of BTERROR.
2587 * In insert and delete operations check for BTLOWER/BTUPPER and isleaf,
2588 if true convert to either slot 0 or Max and perform the insertion.
2589 This produces easier to read code when performing lookups.
2591 * modify lock_AssertXXX macros to call osi_assertx() and provide a
2592 descriptive message.
2594 * the difference between osi_assert() and osi_assertx() is whether or not
2595 a description is specified as part of the call. When no message is
2596 specified afsd_notifier() is called with a NULL msgp parameter. This
2597 results in a null pointer dereference during the EventLog() call.
2599 As a result, none of the information describing the assertion is written
2602 This commit sets a default message for use when no other message is
2603 provided and it replaces all calls to osi_assert with osi_assertx and
2606 * Migrate B+ tree search key into thread local storage
2609 * Fix an uninitialized variable in lana_IsLoopback() which
2610 could result in a random determination that a loopback
2611 adapter is not a loopback adapter. This would result in
2612 a computed Netbios service name of %MACHINE%-AFS instead
2613 of AFS. This is a serious problem if the value computed
2614 by the AFS Client Service and the client process pioctl
2615 calls produce different results as it will result in a
2616 communication failure.
2618 * Fix a crash in the B+ Tree code that results when a file
2619 rename only results in a change of case.
2621 * Fix the pioctl path parsing. The previous algorithm
2622 broke the current directory of the client into two parts,
2623 the TIDpath which represented the device root and the
2624 relative path. The problem is that this is not sufficient
2625 from proper processing of symlinks and mount points.
2627 What is required is that the last component of the path
2628 must not have the symlink or mount point be evaluated
2629 whereas symlink and mount point evaluation must be performed
2630 for the rest of the relative path. Therefore, the path is
2631 now split into three parts: TIDpath, intermediate path,
2634 "fs examine" now reports the data of the symlink or mount
2635 point object instead of the target and "symlink list" and
2636 "symlink remove" now works. The "Explorer Shell Extension"
2637 menu now is displayed for symlink objects.
2639 * More informative symlink.exe error messages.
2641 * When the cmd.exe "del" command is used to remove a symlink,
2642 not only was the link object removed, but the stat cache
2643 object for the target was marked deleted. This resulted
2644 in cache corruption which would require the destruction
2645 of the cache to correct.
2647 * Add undocumented registry value, "OfflineReadOnlyIsValid"
2648 which results in cached contents from read only volumes
2649 to be treated as valid when there is no callback registered
2650 and all servers are down or all instances are offline.
2652 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
2654 * Add documented registry value, "AcceptDottedPrincipalNames"
2655 which is used to permit the acquisition of AFS tokens
2656 using Kerberos v5 principal names which include a dot in
2657 the first component.
2659 {HKLM\HKCU}\Software\OpenAFS\Client
2661 * "fs uuid" does not require "AFS Client Administrator" group
2662 membership to display the current UUID value.
2665 * Update NSIS installer to version 2.30
2667 * Update Wix installer to version 2.0.5325
2669 * 64-bit KFW support:
2670 - Network Identity Manager plug-in
2672 - 64-bit krb5 enable afscreds.exe, afs server manager,
2675 * Enable B+ tree Directory Search by default.
2677 * NIM plug-in is now compatible with FILE: ccaches
2679 * Install instloop.exe into Client/Programs/ even though
2680 it isn't required. Permit it to be used by organizations
2681 that clone Windows systems.
2683 * On a suspend, only drop callbacks for servers that
2684 are online. If the server is offline, maintain the callback
2685 until the server is once again available.
2687 * Prevent against acquisition of a directory write lock when
2688 there are no updates to be performed. This was resulting
2689 in a write-lock leak which later on would lead to a deadlock.
2691 * Add configurable file pre-fetch mode. For files that match
2692 the specified file extensions, pre-fetch the entire file into
2695 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2696 (MULTI_SZ) PrefetchExecutableExtensions
2698 * Fix the client_osi large integer support so that return values
2699 can be properly assigned to large integer variables.
2701 * Do not hold cm_scache_t mutex across cm_GetCell() calls
2702 since cm_GetCell() can block in a RPC.
2704 * Add short file name support to B+ tree searches.
2706 * Resolve a conflict between the desire to notify file servers
2707 that the client workstation is suspending and the desire to
2708 permit objects with valid callbacks when the client suspends
2709 to continue to use those objects from the cache if the client
2710 resumes without network connectivity.
2712 * Fix stack variables to ensure that all buffers storing paths
2713 can handle AFSPATHMAX (1024) characters. Paths of 256 characters
2714 would cause a stack overflow condition.
2716 * On Vista, force afs_config.exe to run with the highest privileges
2717 available to the end user. For a user in the Administrators
2718 group running under UAC, this will cause afs_config.exe to be
2719 be run with full Administrators privileges after the user
2720 approves the privilege escalation.
2722 * More VNOVNODE handling issues. If a dirty buffer cannot be
2723 written due to VNOVNODE, clear the buffer and ignore the error.
2724 If a cm_scache_t object has been marked deleted due to a VNOVNODE
2725 error, the next time the smb_fid_t object that references it is
2726 used return either NOSUCHFILE or NOSUCHPATH and close the
2727 smb_fid_t object. Subsequent attempts to use the smb_fid_t will
2728 result in an invalid handle error.
2730 * For Vista, set the RequestExecutionLevel of the NSIS installer
2733 * For the build environment, support the generation of manifests
2734 when using the CL1310 (VS .NET 2003) compiler/linker.
2736 * There was a bug in OutputDebugHexString() which could overflow
2737 a stack buffer resulting in unexpected process termination
2738 of afsd_service.exe if the input Security Token contained
2739 many bytes equivalent to the '%' char.
2741 * Add directory lookup statistics to "fs memdump" output.
2745 * Windows uses case-insensitive file name pattern matching
2746 but AFS is a case sensitive file system. The AFS3 directory
2747 format is block based, uses network byte order and
2748 includes a hash table for fast case sensitive lookups.
2749 This causes several problems for the Windows AFS client.
2750 (1) Traversing the directory blocks is cpu expensive
2751 (2) A hash table miss does not indicate that the desired
2752 entry does not exist.
2753 (3) Determining whether a non-ambiguous inexact match or
2754 the entry does not exist requires a linear traversal
2755 of the entire directory.
2756 These issues often result in 100% CPU utilization.
2758 These issues are addressed by building a modified B+ tree for
2759 each directory and then using the B+ tree for searches.
2761 This feature is disabled by default. To enable it set
2762 the "BPlusTrees" registry value to 0x01 in the AFS Client
2763 Service Parameters key.
2765 * Correct a null pointer dereference in the symlink recursion
2766 detection code. (Introduced in 1.5.22)
2768 * Add the Network Identity Manager AFS Provider to AMD64
2773 * A new registry value, "BlockSize", can now be used to adjust
2774 the size of the internal buffers used to store file data.
2775 The default is 4KB. When set, the values 1..1024 specify
2776 multiples of 4KB. Values > 1024 specify actual block sizes
2777 and must be multiples of 4KB. The actual block size in KB
2778 can not be larger than the chunk size which defaults to 1MB.
2780 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2783 * pts.exe and ptserver.exe built with -DSUPERGROUPS
2785 * Service Manager shutdown hints have been added. The hint
2786 specifies that it may take the AFS client service up to two
2787 minutes to shutdown.
2789 * Fix the recursion detection feature introduced in 1.5.22.
2790 The previous implementation broke the evaluation of paths
2791 using ../path or ./path. It also broke the evaluation of
2792 symlinks to absolute paths. For example, /afs/cell/path.
2794 * Fix an uninitialized variable in cm_MakeDir that could result
2795 in crash if the requested directory could not be created.
2797 * Fix a the VolumeStatusNotification log messages to save
2798 the input string in the circular log buffer so garbarge
2799 won't be output to afsd.log when the log is dumped.
2802 * Increase the SMB Maximum Path length to 260 characters
2803 from 256 as that is what CIFS supports.
2805 * On 64-bit Windows, use a unique description string for
2806 the 32-bit AFS Shell Extension.
2810 * In the AFS Admin Library, replaced all of the
2811 Ubik_Call(XXX_RPC,...) calls with Ubik_XXX_RPC(...). This
2812 allows parameter checking and proper type conversions to be
2813 performed. This is crucial when passing parameters that are
2814 64-bit wide on systems in which 'long' is a 32-bit value.
2815 Otherwise, the parameter list on the stack is interpretted
2816 incorrectly and the function not only doesn't succeed but it
2817 might corrupt memory and if lucky crash the application.
2819 The only program that we care about that uses this library
2820 is the AFS Server Manager. Now it doesn't crash.
2822 * Fixed smb directory searches for non-wildcard names. The
2823 optimized function should not have set the bulk in progress
2824 flag. This would adversely affect later directory searches
2825 that required a bulk search.
2827 * Detect infinite recursions (up to 512 stat cache objects
2828 in a chain.) If an infinite recursion is detected, return
2829 an error to stop the CIFS client from querying us forever.
2831 * Executables executed out of AFS name space can have their
2832 pages swapped out by Windows. Those pages will be reloaded
2833 when they are required. However, if the cache manager no
2834 longer has a callback and all of the file servers with
2835 the volume data are down, then the application will crash
2836 due to the inability to read the required page.
2838 We now track the down time of each server. If the callback
2839 expires after the server that issued it went down and all
2840 servers are down, we delay the expiration to permit the data
2841 in the cache to be continue to be used by Windows or an
2844 * The Windows AFS client has always had very poor performance
2845 with regards to directory management. In the case of
2846 directory updates caused by create, rename,and delete operations,
2847 the AFS cache manager will now perform local directory updates
2848 when possible in order to avoid re-reading the directory from
2849 the file server. This will be done in whenever both of the
2850 following conditions are true:
2852 - the data version on the directory increased by one as a
2853 result of the operation
2855 - all of the directory buffers necessary to make the change
2856 are present in the cache
2858 * The error message in the AFS Control Panel describing the
2859 need to be a Windows Administrator has been replaced with
2860 a message describing the need to be in the AFS Admins Group.
2862 * If the vldb server list is empty when attempting to evaluate
2863 a cell mountpoint it was possible for a crash to occur. This
2864 can only happen if there is an existing mount point and the
2865 is a failure attempting to load the server list from the
2866 CellServDB file or via AFSDB queries.
2868 * dirty buffers only have their dirty bytes stored to the file
2869 server instead of the entire chunksize
2871 * the default chunksize has been increased from 128KB to 1MB
2876 * changed the enum values for cm_serverRef_t state info to use a
2877 private name space to avoid collisions (srv_)
2879 * added a srv_deleted state for cm_serverRef_t objects. This
2880 state is set when cm_FreeServerList() is called with the
2881 CM_FREESERVERLIST_DELETE flag set. cm_FreeServerList() may
2882 not always delete the cm_serverRef_t from the list if it is
2883 still in use by another thread. the srv_deleted state means
2884 the object's contents are no longer valid and it must be
2885 skipped. It will be deleted the next time the object is
2886 freed and the refcount hits zero.
2888 * the srv_deleted state is also used when a file server reports
2889 either VNOVOL or VMOVED instead of marking the cm_serverRef_t
2890 as offline. This is done to prevent additional usage of the
2891 stale vldb data while waiting for the update volume request
2894 * added a state field to the cm_volume_t object (enum volstate
2895 vl_ name space) that maintains the state of the volume based
2896 upon the states of all of the cm_serverRef_t and cm_server_t
2899 * modified cm_UpdateVolume() to set the state of the cm_volume_t
2900 RW, RO, and BK to either vl_alldown or vl_online. There can't
2901 be any other states because cm_UpdateVolume() destroys any
2902 previous knowledge we might have had regarding busy or offline
2905 * modified cm_UpdateVolume() to update the volume name in the
2906 cm_volume_t to the volume base name if the previous value was
2909 * modified cm_FollowMountPoint() to check to see if the volume
2910 name is a volume ID and if so call cm_GetVolumeByID instead
2911 of cm_GetVolumeByName. This ensures that volume IDs are always
2912 looked up as numeric values. There is no longer a need to
2913 maintain a separate cm_volume_t containing the string representation
2916 * Added a flags parameter to cm_GetVolumeByName() and cm_GetVolumeByID().
2917 The first flag is a "CREATE" flag which is set by all existing
2918 calls. The flag is not set by calls to cm_GetVolumeByID() from
2919 the server probe code when volume status is being updated. We
2920 do not want the server probe operation to result in additional
2921 turnover in the cached volume data. The second flag is NO_LRU_UPDATE
2922 which is set when the server probe code updates the volume status.
2923 This flag will be used to prevent the server probe operation from
2924 changing the order of the least recently used queue.
2926 * Modified cm_GetVolumeByName to ensure that only one cm_volume_t is
2927 allocated for a given set of normal, readonly, and backup volumes
2928 regardless of whether or not the volume is accessed via name or
2929 ID number. The cm_volume_t namep field is always the base name
2932 * Added a new volume state, vl_unknown. This state is used as
2933 the initial state for all cm_volume_t when the cache manager starts,
2934 for each cm_volume_t at creation, and for each cm_volume_t when
2935 recycling. The cache manager does not know the state of all
2936 volumes in the world, only those that are in the cache and for
2937 which it has queried the VLDB and hosting file servers.
2939 * modified cm_GetVolumeByName() to initialize the state of a
2940 volume to vl_unknown. The actual state will be set when a
2941 cm_VolumeUpdate() call completes successfully.
2943 * changed name of scache hash table variables to avoid ambiguity
2944 when adding hash tables for volumes
2946 * fix a buffer overrun in sys\pioctl_nt.c pioctl().
2948 * modified cm_UpdateVolume() to handle the case in which there is
2949 no RW volume but there is are RO volumes for a given base name.
2950 This is done by querying for the ".readonly" volume name if the
2951 base name does not exist in the VLDB. We never query for the
2952 .backup name because under the current usage model a .backup
2953 volume may only exist on the server that the read-write volume
2954 is located. If there is no RW volume, there can be no .backup.
2956 * Added four hash tables for cm_volume_t objects to improve the
2957 search time of cm_GetVolumeByID(), cm_GetVolumeByName() and
2958 cm_ForceUpdateVolume(). One each for Name, RWID, ROID, and
2959 BKID. Three ID hash tables are necessary as long as it is
2960 desireable to maintain a single cm_volume_t containing all
2961 of the related RW, RO, and BK volume data. Having the RW and
2962 RO volume data in the same object is necessary for the
2963 implementation of cm_GetROVolumeID() which returns either the
2964 RO or RW ID depending upon the existence of RO volume instances.
2966 * Added a volume LRU queue so that volume reuse becomes fairer.
2967 This does not replace the all Volumes list which is used when
2968 it is desireable to walk a list of all the volumes whose order
2969 is not going to change out from underneath you which makes it
2970 safe to drop the cm_volumeLock.
2972 * handles volume hash table updates where volume name to
2973 volume ID number changes. The volume name remains
2974 constant in the cm_volume_t. if a vos rename is performed,
2975 the name of the volume will change and the volume IDs will be
2976 updated. Subsequent access to the old volume ID will create a
2977 new cm_volume_t with the new name.
2979 * Added a daemon thread operation to query the state of volumes
2980 listed as busy or offline. cm_CheckBusyVolumes() calls
2981 RXAFS_GetVolumeStatus() for each volume ID that is marked vl_busy
2982 or vl_offline. If the volume is now online, the status on the
2983 volume is updated. The default period is 600 seconds. This can
2984 be configured with the BusyVolumeCheckInterval registry value.
2986 * Added prototype for smb_IoctlPrepareRead() which was missing a
2987 return type in the function definition.
2989 * Added volume id lists to the cm_server_t. These lists are
2990 allocated in blocks of ~32 IDs. When a cm_PingServer()
2991 detects a change in server state, the state of the cm_volume_t
2994 * Added volID to the cm_serverRef_t object. volID is used
2995 to identify the volume for which the object is a referral.
2996 cm_FreeServerList() uses the volID to remove the volume
2997 from the cm_server_t.
2999 * In cm_Analyze, when VNOVOL or VMOVED are received,
3000 call cm_ForceVolumeUpdate() to force a refresh of the volume
3003 * Added cm_VolumeStatusNotification() which is used at the moment
3004 to log volume status changes to the trace log. It will also
3005 be used as the access point to the File System Filter driver
3006 notification engine.
3008 * Added an all cm_scache_t list to cm_data. This replaces the use
3009 of the stat cache LRU queue when we need to enumerate all
3010 entries. The LRU list order is not static and when using it to
3011 enumerate all entries it can result in items being missed or
3012 items being processed more than once.
3014 * Modified cm_Analyze(). Instead of reseting the busy or offline
3015 state of a volume and forcing a retry of the operation
3016 cm_Analyze will defer to the background daemon thread that will
3017 update the state once every 600 seconds.
3019 * Added the automatic generation of a Freelance ".root" read-write
3020 mountpoint that refers to the root.afs volume of the workstation
3021 cellname at the time the mountpoint is created.
3023 * re-write cm_Analyze to make better use of the known volume
3024 status. VL_Server queries cannot result in CM_ERROR_ALLOFFLINE
3027 * renamed cm_CheckBusyVolumes to cm_CheckOfflineVolumes.
3028 busy volumes will be reset to srv_non_busy by the function
3029 but there is no mechanism for querying the busy state other
3030 than by attempting to access the resource.
3032 * cm_Analyze will query the state of an offline volume before
3033 deciding whether or not to retry when all volume instances
3036 * the algorithm used after an IP address list change detection
3037 has been modified. we now check the state of the VLDB servers,
3038 then reset the rx connections, and finally check the state of
3039 the file servers. Previously we flushed the rx connections
3040 and then checked the servers in whatever random order they
3041 were first discovered.
3043 * fix the volume status reported by "fs examine"
3045 * various improvements to the AFS Network Identity Manager Provider.
3046 - Don't access freed objects after the window has been destroyed.
3047 - Write back the AFS cell list in the identity configuration
3048 even when its empty.
3049 - Disable all use of the provider when AFS is disabled for a
3050 particular identity.
3052 * Fix cm_cell_t leak. When evaluating a mountpoint that uses an
3053 alias instead of the full cellname, double check the full cell
3054 name and see if we have a matching cell before allocating a new
3057 * Dump more data with "fs memdump". All volumes and scaches and
3060 * Reverse the order of the all cells list. This ensures that
3061 the workstation cell is always first in the list.
3063 * If we recognize the server from which a callback revoke was
3064 received, then we can ensure that we only invalidate the status
3065 information for cm_scache_t objects belonging to that cell.
3067 * When building for amd64, use APPVER 5.02 as we do not support
3068 64-bit Windows 2000 and explicitly state the machine type in
3071 * no longer treat VMWare network adapters as equivalent to a
3072 loopback adapter. For some unknown reason, publishing the "AFS"
3073 netbios name on them frequently results in a name collision
3076 * Use read locks whenever possible instead of write locks when
3077 walking the allCellsp list. This permits better use of multiple
3080 * In 1.5.20 a bug was introduced by the fake status when PRSFS_READ
3081 is not present on the directory optimization. It would report
3082 mount points and directory objects for which there was no
3083 status information as a zero length file. Instead we can use
3084 the vnode number of the object to determine if it should be
3085 considered to be a directory or a file. If it is odd it is a
3086 directory; even it is a file.
3088 * Give Back Callbacks to the file servers in response to suspend
3089 and shutdown power management events. Only give up callbacks
3090 to UP servers and do so with a very short RX dead timeout.
3093 * Fix a race condition in the SMB FID cleanup code that could
3094 result in a stat cache mutex never being released.
3096 * Fix the graphics files that are installed as part of the
3097 IBM Administrators Guide.
3099 * Change the behavior of pioctl path parsing to not follow
3100 symlinks or mount points during evaluation. This results in
3101 the FID being the symlink or mount point and not the object
3102 they point to. This is consistent with the behavior of the
3105 * Enable AFSDB record lookup for command line tools such as
3108 * CellServDB update 2007-07-05
3110 * Added a Volume Status Plug-in Interface that permits third parties
3111 to receive notifications when the volume state changes occur.
3113 * Ensure that all writes to the afsd_alloc.log file use CR-LF.
3115 * Prevent the afsprocmgmt.dll from being unloaded while the
3116 background thread it starts is running.
3118 * Avoid a race condition when closing smb_fid_t objects.
3121 * fix install path of IBM Quick Start Guide when installed by
3124 * fix the client so that it can write into directories with
3125 only 'li' permissions.
3127 * fix installation of 32-bit Shell Extension for 64-bit systems.
3128 Install it under its own HKCR keys.
3130 * modify the wix installer to remove and upgrade install packages
3131 with the same version number.
3133 * when flushing volume location data every two hours or in
3134 response to "fs checkvolumes", also flush the mount point
3135 links in order to force their re-evaluation.
3137 * aklog was copying an uninitialized buffer as part of the
3138 foreign cell automatic pts registration
3140 * when obtaining directory listings, check for read permission
3141 on the directory. If read permission does not exist, fake the
3142 status information for the directory entries. This improves
3143 performance, prevents the client from pounding the file server,
3144 and prevents leakage of cached status information.
3146 * "fs getcacheparms" output was incorrect.
3148 * NetIdMgr AFS Provider should only automatically add token
3149 acquisition for cell foo.com if the realm of the cell matches
3150 the realm of the identity.
3152 * Fix a null pointer dereference in cm_MergeStatus()
3156 * network identity manager provider:
3158 - Correct for the fact that pr_CreateUser tampers with the
3159 contents of the username input buffer.
3161 - Take into account that the krb4 credential data structure
3162 does not have separate realm fields for the user and service.
3163 When krb524 is used with a cross-realm ticket, the realm
3164 describes the service not the user.
3166 * Many integrated logon changes:
3168 - Fix eventlog reporting. Do not attempt to log an event
3169 if the event source registration fails. Use DebugEvent0
3170 instead of DebugEvent when there are no parameters.
3172 - Modify the LOOKUPKEYCHAIN macro to recognize ERROR_MORE_DATA
3175 - Fix the reading of Domain specific configuration for
3176 LogonScript and TheseCells. Previously the dwSize value was
3177 being overwritten so that subsequent RegQueryValueEx call
3180 - Fix a memory leak in the TheseCells reading code.
3182 - Add support for Domain specific "Realm" specification. The
3183 realm is the realm to be appended to the username. When
3184 logging in as a domain or to the local machine, the specified
3185 "Domain" name is not going to be a valid realm name.
3187 - Construct a proper principal name based upon the domain
3188 specified realm for use in obtaining tokens with KFW.
3190 - If the domain specified "TheseCells" list includes the
3191 default cell, do not obtain tokens twice.
3193 * An unprotected call to the Kerberos v5 function,
3194 krb5_free_context(), can result in an invalid memory access
3195 exception if the contents of the krb5 profile, krb5.ini, cannot
3196 be parsed correctly. The error is located within the afskfw
3197 library which is used by afscreds.exe, afssvrmgr.exe, and
3198 afslogon.dll. When the error is triggered within afslogon.dll
3199 a blue screen condition can occur when the winlogon.exe system
3200 process terminates unexpectedly.
3204 * Do not stop the afs client service with an assert
3205 if the cell name cannot be determined when responding
3206 to a "fs whereis" command.
3208 * Integrated Logon (afslogon.dll). No longer attempt
3209 to preserve Kerberos 5 tickets for the logon session
3210 as that job is now supported by KFW 3.1 and above.
3212 * The Network Identity Manager plug-in will attempt to
3213 remove expired service tickets from the cache.
3217 * CellServDB update - GCO Public CellServDB 09 Mar 2007
3219 * AFS Server Manager and AFS Account Manager data cache
3220 moved from HKCU\Software\TransarcCorporation to
3221 HKCU\Software\OpenAFS to prevent crashes due to data
3224 * Reduce the default UP server probe time from 10 minutes
3225 to 5 minutes to match UNIX. This will not solve the NAT
3226 port mapping timeout issue because most NATs default to
3227 a value less that 5 minutes and this value ensures that
3228 the afs client server will start probes 5 minutes after
3229 the last probe set ended. The probe period for any one
3230 server is therefore longer than 5 minutes. The more
3231 servers in the list the longer the time between probes.
3233 * Before marking a server as down in cm_Analyze, retry the
3234 request using a new rx connection to the server. If the
3235 cache manager is behind a NAT, it is possible that the
3236 response from the server is being sent to the original
3237 address:port from which the rx connection was established.
3238 This will appear to the cache manager as the server being
3239 down when in fact it is not. Forcing the use of a new rx
3240 connection will result in the server responding to the
3241 most recent address:port assigned to the client by the NAT.
3243 If the server is in fact down, it will simply take a little
3244 longer to notice it.
3248 * Removing a directory that contained entries was failing with
3249 the wrong error code being returned to the CIFS client.
3253 * rxdebug exponential back-off. don't take down a server that
3254 is already struggling to respond.
3256 * Network Identity Manager plug-in registers help on the Help
3257 menu if it is supported by NIM. (KFW 3.2 and above)
3259 * Plug-n-play networking did not always work after coming back
3260 from a Windows Suspend. The AFS client service failed to
3261 re-compute the LAN adapter to bind to and instead attempted
3262 to bind to all LAN adapters. This would polute the public
3263 network with the name "AFS" which could lead to conflicts
3266 * Build installers using Microsoft Hotfixes provided at
3267 http://tinyurl.com/243svk for Visual Studio C RunTime Library
3268 Daylight Savings Time issues.
3270 * "fs checkserver" does not require admin privileges unless
3271 the probe interval is being changed.
3275 * A reference leak was discovered in the mountpoint traversal
3276 functionality which resulted in cm_volume_t objects not becoming
3277 eligible for re-use. This will trigger a panic condition when
3278 more than the maximum number of volumes is accessed.
3280 * Fixed the success test for the "fs memdump" operation. Success
3281 is now reported as such. Added cm_volume_t object information
3284 * A data corruption error has been fixed. Write requests
3285 queued for background processing were not retried upon failure.
3286 This leads to inconsistencies between the contents of the file
3287 in the local cache and the contents of the file on the file
3290 The implemented fix is to re-queue any request that has failed
3291 due to timeout, volume busy, volume offline, server busy, or
3292 server offline. Requests are not processed while all servers
3293 for a volume are inaccessible. When servers become accessible
3294 the background request will be retried.
3296 In addition, if a background write fails, do not mark the buffers
3297 as clean. Failure to preserve the dirty flag prevents the buffers
3298 from being written to the file server in the future.
3300 This fix does not address the possibility that the credentials
3301 necessary to perform a write request might expire between the
3302 time the request is queued and when the request can be processed
3303 due to server accessibility.
3305 * The Freelance volume does not support per user ACLs. The
3306 cm_HaveAccessRights function did not handle this special case.
3307 Requests for privileges greater than read|lookup would result
3308 in afsd_service.exe entering an infinite loop. ISOBuster is
3309 known to cause this behavior.
3311 * When deleting files, the SMB Server would attempt to obtain
3312 the access permissions for the parent directory instead of
3313 for the file itself. This would result in an incorrect
3314 assessment of whether or not the file can be deleted.
3316 * The Network Identity Manager AFS plug-in has had its string
3317 tables updated for consistency with the language used by NIM.
3318 All credentials are referred to as credentials instead of tokens.
3319 Drop-down combo boxes now in fact drop down.
3321 * The AFS Salvager was not properly being built on Windows.
3322 This is relevant only if the AFS servers are in use.
3324 * The default signal handlers for the AFS servers incorrectly
3325 processed SIGQUIT. Instead of stopping the server, an
3326 exception was raised.
3328 * Fix another volume refcount leak. This one was triggered by
3329 an inability to contact the vlservers in order to update the
3332 * Fix checklist control used by the drive mappings tabs in
3333 afscreds.exe and afs_control.exe.
3335 * Improve output of "fs memdump" related to buffer management.
3337 * Fix refcount leak of cm_buf_t objects when prefetching buffers
3338 already in the process of being fetched.
3340 * Improve cache performance by computing the buffer hash table
3341 size based upon the number of buffers so that the average
3342 number of entries in a bucket is seven.
3344 * Improve cache performance of read/write data by locally
3345 updating the dataVersion of all buffers associated with the
3346 file on each StoreData or StoreStatus operation provided that
3347 the dataVersion was incremented by one and the dataVersion of
3348 the buffer matched the original dataVersion of the cached
3349 status information for the file.
3351 * When recycling cm_scache_t objects, make sure that we
3352 clear all of the previous fields. It is no longer the
3353 case that we can assume that the stat cache object is being
3354 obtained only after a successful FetchStatus call.
3355 InlineBulkStatus will allocate stat cache objects for which
3356 the no status information is available.
3358 * When a file is being opened with the OPEN_ALWAYS disposition,
3359 do not require write permissions when the file already exists
3360 unless of course the file is being opened requesting write
3363 * When a file has been opened for read-only access, if the
3364 application requests an exclusive lock, obtain a read-lock
3365 instead of a write-lock. The purpose of the exclusive lock
3366 in this situation is to prevent changes to the file and the
3367 full file read lock will provide this guarrantee.
3372 * Enforce Unix Mode Write bit to prevent file deletion of read only
3375 * The CIFS READ and LIST permissions use the same value 0x0001.
3376 When a directory is being accessed, check for PRSFS_LOOKUP
3377 instead of PRSFS_READ.
3381 * Fix computations of short names of mount points. They were
3382 broken as part of the directory search optimization. (1.5.13a)
3384 * When all buffers are in use, periodically sleep the threads
3385 waiting for a buffer to become free while the global buffer
3386 lock is not being held in order to permit those threads that
3387 can free buffers to do so.
3389 * Power Management improvements. Maintain a global flag that
3390 specifies whether or not the service is in a suspend state.
3391 Do not panic if Netbios() returns NRC_BRIDGE meaning that the
3392 lana is no longer valid. Instead, stop the listener threads
3393 and if all listener threads are stopped, reset the lana_list.
3394 Allow the cm_Daemon() thread to periodically check the state
3395 of the smb listeners. If they are all stopped and the service
3396 is not suspended, attempt to restart them. If there are no valid
3397 lanas, return to the stopped state.
3399 * CreateX and NTCreateX use cm_CheckNTOpen() to test whether or not
3400 the user is permitted to obtain read or write locks. This function
3401 would obtain the lock and then drop it returning whether or not
3402 the lock could be obtained. If the lock was in fact required,
3403 CreateX/NTCreateX would then obtain it with cm_Lock(). The problem
3404 of course being that this pattern results in three RPCs to the
3405 file server (lock, unlock, lock). This is reduced to one RPC by
3406 implementing cm_CheckNTOpenDone() which frees the allocated byte
3407 range lock from cm_CheckNTOpen() after the cm_Lock() call is
3410 * Optimize the SMB FindFirst processing for the case in which there
3411 are no wildcards. If a cm_Lookup succeeds, generate a response
3412 from the stat cache entry we have a callback for. Its cheaper
3413 than walking each buffer of directory data associated with the
3414 directory stat cache entry when we have thousands of directory
3417 * Increase the SMB Ioctl MaxData size to support larger token sizes
3419 * Disable autorun of the AFS Server Wizard when the AFS Server is
3422 * Add AFS Server service dependencies on tcpip and PNP_TDI
3424 * Remove the AFS Server Configuration Tool shortcut
3426 * Add Freelance mode awareness to AFS Server Configuration Tool
3430 * Unlike previous versions of the OS, Vista performs a shutdown on
3431 the Microsoft Loopback adapter just like it would on a real adapter.
3432 This causes the smb_Listener threads to trigger a panic during a
3433 suspend/hibernate power event.
3435 The fix is to unbind from the network adapters in response to a
3436 suspend/hibernate power event and then rebind to the adapters
3437 when a resume power event is received. Note that the resume events
3438 are not reliably delivered so it is possible the afs service will
3439 not be accessible. However, this is the best we can do.
3441 * don't allow environment variables we set in afslogon.dll to be
3442 inherited by child processes
3444 * force an rx connection reset if VICECONNBAD or VICETOKENDEAD are
3447 * The old dirty buffer synchronization algorithm had a buf_IncrSyncer
3448 thread walking the all buffer list periodically searching for dirty
3449 buffers to write to the file server. This had several negative
3450 results. The algorithm ate up ever increasing amounts of CPU time
3451 even when AFS is idle as the size of the cache increases. Also,
3452 buffers were written to the file server in an order based upon the
3453 original buffer allocation which has nothing to do with the order
3454 in which the buffers became dirty.
3456 The new algorithm maintains a dirty buffer list. Items are added
3457 when the buffer is originally marked dirty. A buffer is only
3458 removed from the list by the buf_IncrSyncer when the buffer is no
3459 longer dirty. If the list is empty the thread goes back to sleep
3460 immediately without additional processing requirements.
3464 * Enable renaming of directories from the Explorer Shell when
3465 Offline Folders is enabled
3467 * An error is reported in response to the SMB ReceiveCoreFlush
3468 request when an error has not occurred.
3472 * optimize dirty buffer flushes in response to suspend operations
3473 do not flush if only loopback adapter is installed
3474 do no retry dirty buffer writes
3476 * do not reset netbios when performing loopback checks. doing
3477 so will break the existing lana bind
3479 * remove a deadlock in smb3.c QPathInfo
3483 * Add NetIDMgr AFS Plug-in
3485 * "Show Tray Icon" checkbox removed from afs_config.exe
3487 * Disable DNS Registrations for the Loopback Adapter and make sure
3488 that Netbios is turned on.
3490 * A reference count leak on stat cache objects was fixed
3492 * For debugging, tread ids of writers are now tracked on locks
3494 * Corrected an incorrect lock that was being held but not
3495 freed in a rare error case
3497 * Changed trace log end of line to CR-LF so it can be read
3500 * modify the buffer synchronization thread to write all dirty
3501 buffers and sleep only when there is no work to be done
3503 * improve behavior of cache manager when large numbers of
3504 temporary files are being created by multiple clients within
3507 * do not permit dirty buffers that belong to deleted files
3508 to be written to the file server. its a waste of bandwidth
3510 * do not leak stat cache references when the file server
3513 * afscreds: do not permit an attempt to obtain tokens if the
3514 afs service could not be started. this was causing stack
3515 overflows on Vista after resuming from a suspend
3519 * better handling of dot directory
3521 * fs sq and fs mkm now behave as on UNIX
3523 * add license to installers
3525 * Return Invalid Parameter error for unknown InfoLevels in
3526 SMB Tran2 Search Directory.
3528 * Correctly use WSAStartup and WSACleanup within the RX library
3530 * Check that afsd_service.exe is running before performing
3533 * Force AFS Shell Extension to load resource DLL using
3534 afs_shl_ext.dll as a base name.
3536 * (Wix) install afs_shl_ext_XXXX.dll and afs_creds_XXXX.dll in
3537 OpenAFS\Client\Program
3539 * Fix SMB_SET_FILE_END_OF_FILE_LENGTH
3541 * Do not panic if the maximum number of volume entries are in use
3542 and one of them can be recycled.
3546 * Fix the code that reads the backconnnectionhostnames value
3547 so that there is not an uninitialized variable
3551 * integrated logon could cause winlogin.exe to blue screen windows
3552 if it was called with a domain that does not exist on the machine.
3553 this is possible during a sysprep.
3555 * corrections to cm_buf_t handling to ensure that buffers are
3556 removed from queues before they are freed and that holds are
3557 obtained on the next buffer in a chain before releasing the
3560 * do not generate MessageBox dialogs on network errors. they
3561 cause more trouble then they are worth.
3563 * move GlobalAutoMapper into a separate thread so that it does
3564 not prevent service startup if the WNetAddConnection2 call
3567 * restrict hard dead and connection timeouts to the CIFS Session
3568 Timeout value. This is being done to try to prevent virtual
3569 circuits from being destroyed prematurely.
3571 * Fix the code that writes the backconnectionhostnames value
3572 to ensure that the data buffer is written with the correct
3575 (there were no windows specific changes between 1.5.4 and 1.5.5)
3579 * Corrections to the Integrated Logon functionality that permits
3580 Kerberos 5 TGTs to be communicated into the user session. The
3581 transfer is now more secure and succeeds when the user does not
3582 have read permission to the %SystemRoot%\TEMP directory.
3584 * NSIS installer updated to version 2.18
3586 * WiX installer updated to version 2.0.4221.0
3588 * Updates to the IFS functionality for 64-bit correctness and
3589 permit compilation. The IFS functionality works for pioctl()
3590 operations but not much else at this point.
3592 * A major re-write of the OpenAFS implementation of the CIFS Tran2
3593 query and set functions.
3595 * Ignore the Windows CIFS client session timeout entirely. Timeout
3596 only after the AFS Hard Dead Timeout period expires.
3600 * move headers, libraries, and samples from OpenAFS\Client to
3603 * remove the auto-registration of AFS IDs for foreign cells
3604 from integrated logon. this permits afslogon.dll to avoid
3605 initializing the RX library. afslogon.dll is a network
3606 provider library and as such it is loaded into all applications
3607 which perform a CreateProcessAsUser call. Initializing the
3608 RX library spawns background threads to listen for inbound
3609 packets. When the network provider is unloaded from the
3610 process, the RX library does not cleanup the threads. If they
3611 continue to execute after the library is unloaded, the process
3614 * implement inline bulk status calls. this will significantly
3615 improve the performance of directory listings.
3617 * prevent a crash when evaluating mountpoints to volumes that do not
3620 * new command: fs uuid [-generate]
3621 displays the current uuid and allows a new one to be generated
3622 on the fly without restarting the afs client service.
3624 * fix the 64-bit installers to properly document the install
3627 * move help files to the proper location so they can be located by
3628 the system tray tool and the control panel.
3630 * "over quota" and "disk full" errors were not being reported to the
3633 * largefile support has been added. Files greater than 31-bits can no
3634 longer be written to file servers which do not support large files.
3635 When file servers support large files, files up to 64-bits can be
3638 * cm_SyncOp could return with the input buffer unlocked upon error
3639 which would result in a panic condition. This bug was introduced
3642 * vlserver failover when freelance mode did not work. This prevented
3643 the root.afs volume from being loaded if the first vlserver could
3648 * A bug in the CIFS Standard Info response packet caused directories
3649 to appear as standard files. This prevented Microsoft Vista from
3650 being able to walk the AFS directory tree.
3652 * Mount points must be reported as directories in response to
3653 Query Path Info operations.
3655 * A deadlock was detected when performing "fs flushall" if the
3656 file server reports VNOVNODE. The scp->createBufferLock is
3657 already held by the current thread and there is no mechanism
3658 to propagate the knowledge. Therefore, an alternate mechanism
3659 for clearing the cache must be developed. A new function
3660 cm_RecycleSCache(scp,flags) has been extracted from
3661 cm_GetNewSCache(). This function performs the task of recycling
3662 an cm_scache_t object. When called from cm_GetNewSCache() with
3663 no flags the expectation is that there are no associated buffers
3664 that are queued to be read or written. When called from
3665 cm_Analyze() with the CM_SCACHE_RECYCLEFLAG_DESTROY_BUFFERS
3666 flag, any queued buffers will be de-queued and marked as if
3667 the operations were performed so the data can be discarded.
3669 * CM_ERROR codes were logged as "unknown" by cm_Analyze.
3671 * In response to VNOVNODE, the parent is only discarded if the current
3672 cm_scache_t is not a directory.
3674 * In the Ioctl FlushFile and FlushVolume functions, there were no
3675 checks to protect against flushing the Freelance SCache entries.
3677 * In FlushFile, the wrong cm_scache_t object was being released.
3679 * In cm_GetNewSCache, do not allow Freelance SCache entries to be
3680 recycled. Choose a new entry if cm_RecycleSCache fails.
3682 * Speed up the performance of the cache manager by not
3683 holding the smb_fid_t mutex across calls to cm_SyncOp
3686 * Ensure that all smb_fid_t flag references are protected.
3688 * Remove a deadlock between smb_fid_t mutex and
3689 smb_rctLock introduced in 1.4.1.
3691 * Fix aklog AFSID lookup to use not send the realm
3692 name when the realm is the local realm for the
3693 cell. The PT Server doesn't strip the realm
3694 name when it is the local realm.
3696 * Treat "\\afs\*." as an alias for "\\afs\all"
3698 * Dynamically adjust the priority of server threads
3699 based upon the age of the CIFS request that is
3700 being processed. The priority is bumped one level
3701 for every 15 seconds of processing time.
3703 * The original openafs contribution mis-used the
3704 cm_GetCallback function. This function should only
3705 be called as a side effect of a call to
3706 cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK). The way it
3707 was being called results in a FetchStatus RPC being
3708 performed even when the client already has a valid
3709 callback and does so in a manner that destroys the
3710 synchronization of the threads that are calling
3711 cm_SyncOp. If nothing else this results in a
3712 significant performance penalty.
3714 * More changes to cleanup of smb_vc_t objects in
3715 order to prevent race conditions.
3717 * Reduce the amount of Power Management event logging
3720 * If the client detects that the IP address has
3721 changed, force the use of new RX connections. This
3722 enables the file server to recognize that the client
3725 * Remove a race condition and deadlock associated
3726 with cleanup of smb_vc_t objects.
3728 * Fix the ANSI filename support. One function that
3729 required it was missing the translations.
3731 Changes not present in 1.4.1 and since 1.5.0: 1.5.1 released
3734 * Add a DllMain() function to afspthread.dll. Perform
3735 cleanup of thread local storage, rx queues, critical
3736 sections, event handles, etc. This was implemented
3737 to avoid crashes in afspthread.dll when applications
3738 which load/unload the dll attempt to terminate.
3740 * Lock request behavior has been altered. As of this
3741 release a locally allocated lock will be used whenever
3742 the volume is RO or the requesting user's permissions
3743 are 'rl'. This will allow end users to execute programs
3744 from RW volumes when they do not have the 'k' privilege.
3746 Note that a recent discovery indicates that the 'k' bit
3747 is only supposed to control the ability to obtain read
3748 locks. Write locks are to be controlled by the 'w' and
3751 * Code has been added to enable the cache manager to be
3752 able to obtain the AFSID of the authenticated user.
3753 This is required to be able to determine when "Insert"
3754 privileges should apply. AFS considers a file to be
3755 a new file whenever the owner's AFSID matches the
3756 AFSID of the user. The current implementation relies
3757 on the name specified during the SetToken call to be
3758 correct. This will be replaced in the future with a
3759 new RPC that allows the AFSID of the authenticated
3760 user to be obtained.
3762 * A 32-bit-tools installer for use on 64-bit systems
3763 is now available. These tools are for use with 32-bit
3764 versions of Kerberos for Windows or JPSoftware's 4NT.
3766 Since 1.4.1 rc10: 1.4.1 final released (9 April 2006)
3768 * Remove the use of the "active_vcp" variable. This variable
3769 no longer serves any useful purpose and was the cause of
3770 race conditions that produced under counts of the smb_vc_t
3773 * Change default minidump output to include data segments
3775 * restore access to cm_IoctlGetVolumeStatus and
3776 cm_IoctlSetVolumeStatus calls for AFS pioctl operations.
3777 Still need to figure out why they are being called
3778 for SMB/CIFS IOCTL operations instead of Trans2GetFileInfo.
3780 Changes not present in 1.4.1: 1.5.0 released (16 Feb 2006)
3782 * Update build system to include the "sys_name" in the
3785 * optimize the number of calls to obtain the smb_user_t for a
3786 smb virtual circuit.
3788 * add logging of server addresses, status and capabilities when
3791 * prevent multiple threads from pinging the same server simultaneously.
3792 instead, threads block and then use the result of the ping that was
3793 already taking place.
3795 * Add registry values to allow the defaults for daemon checks to be altered
3796 at startup. Log values to afsd_init.log.
3798 daemonCheckDownInterval
3799 daemonCheckUpInterval
3800 daemonCheckVolInterval
3801 daemonCheckCBInterval
3802 daemonCheckLockInterval
3803 daemonCheckTokenInterval
3805 * Add EventLog messages for transitions in fs crypt state.
3807 * Byte range locking (when locks will be obtained from the
3808 is still under discussion)
3810 * 64-bit Windows operating systems now supported
3811 (XP SP2 64-bit, 2003 SP1 64-bit, 2003 R2 64-bit)
3812 This is a native 64-bit implementation of OpenAFS for Windows.
3813 There is currently no 64-bit version of KFW and therefore
3814 there is no (easy) method for obtaining AFS credentials.
3815 32-bit KFW 3.0 can be used with 32-bit OAFW libraries
3816 to obtain AFS tokens but there is not for the feint of
3819 * A faster implementation of the fcrypt algorithms improves
3822 * The afsd_service.exe now supplies Event Log Message Text.
3823 The event log no longer generates warning text due to lack
3824 of message configuration.
3826 * When afs servers are unreachable and network stack returns
3827 a WSAEHOSTUNREACHABLE error, the AFS client will immediately
3828 mark the attempted server as being down. This allows for
3829 very fast response to the calling application. This change
3830 has a pronounced improvement when the Windows Explorer is
3831 used on disconnected mobile computers.
3833 * Added DebugOutputString debugging of the RX library
3835 Since 1.4.1 rc9: 1.4.1 rc10 released (10 March 2006)
3837 * Fix the rx-lwp implementation to randomly generate rx call
3838 identifiers which are supposed to be unique at a given point
3839 in time. This affects some of the command line tools used
3840 for debugging but not the AFS Client Service itself which
3841 uses the rx-pthreads implementation.
3843 * xstats_fs_test.exe and xstats_cm_test.exe are now built for
3844 Windows although they are not distributed
3846 * don't generate an exception when logging unexpected SMB packets
3848 * the data structures used by cm_IoctlGetVolumeStatus and
3849 cm_IoctlSetVolumeStatus are not compatible with the CIFS
3850 Get/Set File Information calls so don't bother executing them.
3854 * Fix data written to registry as part of BackConnectionHostnames
3857 * Fix initialization bug when allocating cm_server_t objects that
3858 would cause servers to appear to be down after each allocation
3859 or renewal. Server would appear up after the next check servers
3864 * Windows will return WSAECONNRESET to the next receive packet
3865 operation on a socket if an ICMP packet is received in response
3866 to a packet that was sent and could not be delivered because
3867 the port was unreachable. We have to treat this as
3868 WSAEWOULDBLOCK so that we can retry the select() check and
3869 not mark the connection as bad.
3871 Since 1.4.1 rc6: 1.4.1 rc7 released (14 Feb 2006)
3873 * The client manager maintains a least recently used list of
3874 all the ACLs. If an ACL is already the most recently used
3875 do not bother to remove it from the list and place it on the
3878 * As part of the process of adding a new server, set the
3879 server is down flag and then perform a ping. If the server
3880 is up, the flag will be reset. If the server is down, the
3881 client will not timeout unnecessarily while attempting to
3884 * If a Mountpoint String is the empty string, return PATH NOT
3887 * Modify behavior of the Network Provider to avoid querying the
3888 profile location if integrated logon is known to be disabled.
3890 * Correct behavior of 'LogoffPreserveTokens'. When set, tokens
3891 will always be preserved regardless of whether or not integrated
3892 logon was used to obtain tokens. When unset, tokens will be
3893 destroyed regardless of whether integrated logon was used to
3894 obtain the tokens unless the profile is stored in AFS.
3896 * Disable LogoffTransferToken functionality. The SMB V3 Logoff
3897 message is being delivered even when the user is not logging
3898 off of the computer.
3900 * Fix a deadlock condition introduced in rc6.
3902 * Cell names are case-insensitive. Since the founding of OpenAFS
3903 there has been a bug that treated "cell", "Cell", and "CELL" as
3904 different cells. This would result in what appeared to be
3905 random disappearance of user tokens or a failure to authenticate
3906 even when tokens were listed by the "tokens" command.
3908 Since 1.4.1 rc5: 1.4.1 rc6 released (2 Feb 2006)
3910 * Additional auditing of the smb code revealed that smb ioctl
3911 operations were failing to free smb_fid_t objects
3913 * The File Identifiers assigned to smb_fid_t objects could be
3914 given the value 65535 which is 0xFFFF which is also the
3915 INVALID_HANDLE value. Both 0 and 65535 are now treated as
3918 * Audited all assignments to fields and ensured that they are
3919 protected by obtaining the relevant mutex object.
3921 * Re-wrote the VNOVNODE handling to avoid a potential deadlock
3922 if both the current cm_scache_t object and its parent are
3923 both returning VNOVNODE.
3925 * Removed all the logic associated with the 'dead_vcp' variable
3926 which must have been added for debugging because its used
3927 was never thread safe and it truly served no purpose.
3929 Since 1.4.1 rc4: 1.4.1 rc5 released (25 Jan 2006)
3931 * Re-enabled LogoffTokenTransfer and LogoffTokenTransferTimeout
3932 registry settings. Default is ON and 120 seconds. If set to
3933 OFF, tokens will not be destroyed.
3935 * Fixed a number of race conditions and reference count problems
3936 associated with global queues.
3938 * Added a method to detect abandoned smb virtual circuits so their
3939 associated file handles, locks, user tokens, and smb session
3940 resources can be released.
3942 * Added a mechanism to allow user tokens to be preserved
3943 during integrated login and be freed after logoff.
3945 * Check access permissions to the file before the file is
3948 * More improvements to the handling of SMB Virtual Circuits
3949 when SMB sessions are closed.
3951 * Correct a race condition introduced in rc4 that could result
3952 in a crash of afsd_service.exe.
3954 Since 1.4.1 rc3: 1.4.1 rc4 released (11 Jan 2006)
3956 * Properly handle pre-mature SMB session closures.
3957 File handles, tree connections, and user sessions are now
3958 freed when the virtual circuit is closed.
3960 * smb_username_t and smb_user_t objects are now properly reference
3961 counted and freed as appropriate.
3963 * disable to use of AFS file locks. byte range locking is still
3964 implemented but all locks are managed locally. This prevents
3965 multiple processes from editing the same file on the same
3966 machine but it does prevent multiple processes on separate
3967 machines from modifying the file.
3969 Since 1.4.1 rc2: 1.4.1 RC3 released (4 Jan 2006)
3971 * A fix to "fs setserverprefs -vlserver". Multiple calls with
3972 the same server parameter would result in a panic condition
3973 if the server had not already been contacted by the AFS
3976 * Better warnings are written to the afsd.log file when
3977 the Windows RPC Protocols are not properly configured as
3978 is often the case on Windows 2000.
3980 * fix the procmgmt library to restore C RTL signal handlers when
3981 it is being detached from the process prior to process termination.
3983 Since 1.4.1 rc1: 1.4.1 RC2 released (30 Nov 2005)
3985 * close a lock leak in the rx library that could be triggered while
3988 * prevent removal of drive mappings in response to afscreds.exe -M
3990 * add logic to process VNOVNODE in cm_Analyze. Force re-evaluation of symlink
3991 strings and flush the stat cache entry.
3993 * prevent server objects from being freed if user preferences are set.
3995 * Update WIX support to 2.0.2419.0
3997 * Release all locks on the server at service shutdown.
3999 * A down server should be marked up as long as it responds to the client.
4000 It doesn't matter if the server returned an error as long as the error
4001 is something other than VRESTARTING. If there is such an error, all
4002 connections on that server should be marked for replacement.
4004 * Change the check "up" servers period from once per hour to once every
4005 ten minutes. This matches the behavior of the Unix cache manager.
4007 Post 1.4.0 revisions: 1.4.1 RC1 released (21 Oct 2005)
4008 * Byte Range Locking support has been added. Architectural documentation
4009 is located within src/WINNT/afsd/cm_vnodeops.c. Best efforts are
4010 made to enforce Windows mandatory locking model using locally managed
4011 byte range locks with full file locks obtained from the file server.
4012 Read-only volumes provide the user with an implied locking privilege.
4013 Read-write volumes require that a user be granted the locking privilege
4014 in order for a lock to be obtained. As most Windows applications use
4015 Shared file access modes, they require the ability to obtain locks in
4016 order to access files.
4018 * Implement error translation functions to convert Universal Error Codes
4019 output by modern servers to local errno and winsock error values.
4020 This fixes the error message problems using pioctl() calls from fs.exe
4021 and the AFS Shell Extension.
4023 Since 1.4.0 rc8: 1.4.0 final tagged (20 Oct 2005); released (31 Oct 2005)
4024 * Allow the AFS Client Service to set the firewall configuration by
4025 retrying until after Windows determines the boot period is over.
4027 * Do not attempt to configure the firewall on Windows 2000, XP pre-SP2,
4030 * Detect IP address changes from within the AFS Client Service for the
4031 purpose of checking down servers sooner
4033 * If the file server or vldb server report RXKADUNKNOWNKEY, return
4034 SEC_E_NO_KERB_KEY to Windows. This provides an indication to the
4035 user that there is a Kerberos key problem. Previously the user would
4036 receive no feedback when there was something wrong.
4038 * Fix the large integer library support to properly handle values greater
4041 Since 1.4.0 rc7: 1.4.0 rc8 released (15 Oct 2005)
4042 * Updated CellServDB
4044 * Fixed file name pattern matching. Question marks can match nothing
4045 if found at the end of a component.
4047 Since 1.4.0 rc6: 1.4.0 rc7 released (8 Oct 2005)
4048 * Kerberos 5 ticket lifetimes should adhere to KFW Leash lifetimes
4050 * The MSI installer now has a new Product ID for each build. This
4051 is to allow for automated upgrades. The Upgrade ID is remaining
4054 * Volumes that were cached and then subsequently were salvaged,
4055 moved or otherwise taken temporarily offline (as opposed to being
4056 busy) would become unreachable. A distinction is now made between
4057 a volume being offline and a server being down.
4059 Since 1.4.0 rc5: 1.4.0 rc6 released (2 Oct 2005)
4061 * Fixed several AFS Shell Extension issues:
4062 - Checking Server Status would only work if a specific cell was
4064 - If servers were down, the Server Down dialog would write to
4065 invalid memory and cause explorer.exe to crash
4066 - Default Push Buttons were not specified for many dialogs
4067 - The extension would be unloaded by the Explorer Shell and
4068 never be reloaded under some circumstances.
4070 * Fixed the AFS System Tray menu to disappear if another Window
4071 was selected before a menu item was selected
4073 * Added volume owner and group information to "fs examine" output
4075 * Added fs and registry support for enabling/disabling/clearing
4076 rx statistics gathering.
4078 Since 1.4.0 rc4: 1.4.0 rc5 released (25 Sep 2005)
4080 * New Product Code GUIDs issues for 1.4 and 1.5 branches. Separate
4081 GUIDs for each platform.
4083 * Documentation directory structure and default installation rules
4086 * New OpenAFS for Windows HTML Release Notes installed
4087 as part of the documentation set.
4089 * Add thread id to rx debugging messages
4091 * When all servers are offline, return a bad network path error
4092 immediately. This is necessary to allow Explorer to do the
4093 right thing and not hang. The background daemon thread will
4094 check the down servers every 30 seconds to see if they are
4097 Since 1.4.0 rc3: 1.4.0 rc4 released (14 Sep 2005)
4099 * The AFSCache file has been modified to store both the
4100 serial number of the volume on which the file resides
4101 and the SID of the local machine. If either change,
4102 a new UUID is generated.
4104 * add "fs flushall" command that forces all data buffers
4107 Since 1.4.0 rc2: 1.4.0 rc3 released (4 Sep 2005)
4109 * add support for '/' instead of '\' in pioctl() calls
4111 * Apply AFS Client Admins group protect to AFS Shell Extension
4113 * Add support for \\afs\<name> to most AFS Shell Extension
4114 commands except the symlink methods.
4116 * when installing the Microsoft Loopback Adapter, enable
4117 MS Client for Networks and prevent an install failure
4118 by not calling CoInitialize twice in the same thread.
4120 * reload cell vldb values from the CellServDB every two
4121 hours in case it changes
4123 * When updating cell information from DNS, be sure to set
4126 * Add support to allow use of \\AFS\<foo> where <foo> is
4127 either a mount point or symlink. As <foo> is normally
4128 treated as a share name, we transform it into \\AFS\all\<foo>
4131 * Init 'code' to prevent false errors when integrated logon
4132 is disabled and the service is not running
4134 Since 1.4.0 rc1: 1.4.0 rc2 released (28 Aug 2005)
4136 * Removed trace log messages that were hampering performance.
4138 * Fixed a deadlock that was being triggered by editing Word
4139 documents stored within AFS with WinWord 2003.
4141 * Bit 3 of the TraceOption registry value is now used to set the
4142 default for "fs trace" in the non-Debug builds. The new default
4143 is off for release builds and on for debug builds.
4145 * Bit 2 of the TraceOption registry value can be used to turn on
4146 real-time output of debug log entries to the Windows Debug
4147 Output monitor interface. This data can be viewed with tools
4148 such as Sysinternal's DbgView.
4150 Since 1.3.87: 1.4.0 rc1 released (17 Aug 2005)
4151 * When the cache manager reports ALLBUSY or ALLOFFLINE for
4152 a resource we should not be returning status codes such
4153 as Network Busy or Remote Host Down because these errors
4154 will imply that the AFS Client Service is Busy or Down
4155 and that is simply not the case. Instead we will return
4156 Bad Network Path as the path is temporarily not available.
4158 Instead of returning Sharing Paused when there is a
4159 timeout error, return Timeout. Once again, the AFS Client
4160 Service is not paused.
4162 * afscreds.exe would display an Obtain Creds dialog when
4163 the expired credentials reminder was triggered even
4164 if there was no network path to the KDC. This is prevented
4165 by adding KDC probe logic to the reminder thread.
4167 * afscreds.exe would display expired tokens no differently
4168 than unexpired ones. This would make it difficult for the
4169 user to distiguish when the tokens were expired. For the
4170 English build added a new resource string "(expired) that
4171 is displayed instead of the expiration time.
4174 * "fs wscell" when executed in freelance mode will return the
4175 name of the cell configured in the registry. The root.afs
4176 volume is not loaded from this cell, but it is used for the
4177 default for aklog and integrated login.
4179 * "fs mkmount, fs rmmount, symlink make, symlink remove" will
4180 no longer work on \\AFS\all when freelance is being used
4181 unless the user is a member of the "AFS Client Admins" Windows
4184 * some more checks performed during persistent cache validation.
4187 * Optimize calls to cm_CheckServer from cm_Analyze. Only check
4188 down servers for the current cell not all cells.
4190 * If the scache LRU list becomes corrupted, the AFS Client Service
4191 might enter an infinite loop at startup. Detect the corruption
4192 and discard the cache contents in this case.
4194 * Fixed the Explorer Shell's remove mountpoint functionality.
4196 * Fixed a deadlock caused by the holder of cm_aclLock attempting to
4197 obtain a mutex lock on a cm_scache_t object whose lock is held by
4198 a thread attempting to obtain the cm_aclLock.
4201 * Really fixed DNS AFSDB queries to ensure that "csail" cannot be
4202 misinterpretted as "csail.mit.edu" when the resolver
4203 is configured to append ".mit.edu" to failed DNS queries.
4205 * Added a new registry key, "LogoffPreserveTokens" (see registry.txt),
4206 that can be used to force the preservation of user tokens upon logout.
4208 * Update the NSIS install scripts to use NSIS 2.07. This release adds
4209 recommended installation categories: AFS Client, AFS Administrator,
4210 AFS Server, AFS Development Kit. Each category includes a different
4211 default subset of the OpenAFS components.
4213 The OpenAFS logo is now associated with the NSIS uninstall entry
4214 in the Add/Remove Programs control panel.
4216 * The user name associated with AFS tokens when obtained with
4217 integrated login, the afs systray tool, or aklog will always
4218 include the full Kerberos 5 user principal name regardless of
4219 whether or not the cell is local to the realm.
4221 * Modify integrated login so that it does not enter an infinite
4222 loop if the service is not set to auto start.
4224 * Added asetkey.exe used to set a Kerberos 5 key for use by the
4227 * Added uninstall.exe to Wix installation
4229 * More modifications to algorithms used to wake sleeping threads.
4233 * Changes to the algorithms used to wake threads when they are
4234 put to sleep because of conflicts over buffer operations.
4237 * The OpenAFS integrated logon obtains Kerberos 5 tickets as part of
4238 the process of producing AFS tokens. The tokens are stored within
4239 the AFS Client Service but the Kerberos 5 tickets have been discarded.
4240 New functionality has been added to temporarily copy the tickets to
4241 a file ccache during the short window between the NPLogonNotify and
4242 WinLogon Logon Event. The Logon Event handler starts a process as
4243 user, afscpcc.exe, that copies the tickets into the default ccache.
4244 The file ccache is then destroyed by the event handler.
4246 * During synchronization operations on stat cache entries, if threads
4247 are forced to sleep it was possible that the threads would not be
4248 woken under heavy load.
4250 * Enforce the delete on close flag specified by the NTCreateX()
4253 * A race condition in the rx library was fixed that resulted in
4254 a deadlock in rx_NewCall when the RX_CONN_MAKECALL_WAITING flag
4255 was cleared when threads were still waiting to use the
4256 connection. Also, fixed a potential case in which the
4257 connection could be destroyed while threads were waiting to use it.
4259 * The DNLC freelist has been seen to become corrupted with still
4260 active entries being stored on the freelist. Changes were made
4261 to perform a better job of cleaning entries before placing them
4262 onto the freelist; marking them with the correct magic value when
4263 purging the list; and allowing the list to be purged on startup
4264 if the cache validation fails.
4266 * Windows Crash Reporting does not get triggered for afsd_service.exe
4267 because it provides its own top level exception handler. This patch
4268 places an explicit call to ReportCrash() on platforms which support
4271 If you configure Crash Reporting via AD Policy to capture crash
4272 reports within the domain, then you will receive this output.
4274 * Generate MiniDumps independent of the Windows Crash Reporting.
4276 (a) If an exception occurs,a minidump will be generated at the path
4277 %TEMP%\afsd.dmp. The type of minidump defaults to Normal but
4278 can be set to any other type via use of the
4281 [HKLM\SOFTWARE\OpenAFS\Client] MiniDumpType (DWORD)
4283 (b) MiniDumps can be generated at any time using the "fs minidump"
4284 command. This allows you to generate a minidump without
4285 stopping the process or even requiring the use of a debugger
4288 * an RX connection which reports bad ticket now treats it the same way
4289 the expired ticket is treated. The ticket is removed from the
4290 connection and a new anonymous connection is established.
4292 * fs memdump now requires that the caller be in the AFS Client Admins
4295 * additional debugging information is provided in cm_Analyze when the
4296 error code is ignored.
4298 * Fix a regression introduced into the pioctl() client code which
4299 will cause a crash in winlogon.exe, explorer.exe and fs.exe if
4300 MIT KFW is not installed. (1.3.82a)
4302 * Remove AFS Gateway option from OpenAFS Control Panel (1.3.82a)
4305 * Fix a race condition in cm_EndCallbackGrantingCall() that
4306 could leave 'nrevp' pointing to freed memory. In the process,
4307 optimize the processing to ensure that the applications monitoring
4308 the status of the stat cache entry are only notified at most
4311 * Conditionalize access to the Explorer Shell AFS->Mount Point
4312 ->Remove menu item. Disable the item if the selected item(s)
4313 are not mount points.
4315 * Activate AFS RPC (RX) Free Packet Queues stored in Thread
4316 Local Storage. This eliminates thread contention on the
4317 global RX free packet queue. This should improve performance
4318 on multi-CPU systems.
4320 * Fix the IP address filtering code to properly load/unload
4321 the IPHelper DLL so that Windows XP/2003 and beyond systems
4322 do not have to manually probe the registry.
4324 * CleanupACLEnt() was not being called consistently with
4325 the cm_scache_t object referred to by the back pointer
4326 mutex locked. This could in very rare conditions lead to
4327 an invalid memory access.
4329 * Added a script command to the msi installer "afs_replace"
4330 which can be used to replace one version of OpenAFS with
4331 another without requiring an uninstall and reinstall if
4332 the installed msi is no longer in its original location.
4334 * Apparently the problem with multi-domain forests with cross-
4335 realm trusts to non-Windows realms was not entirely solved.
4336 The authentication to the AFS SMB service failed because
4337 the wrong name was being used. Using ASU as an example,
4338 the authentication was being performed with the name
4339 "QAAD\user" (an account in the forest root) and not
4340 "user@ASU.EDU (the MIT Kerberos principal used to login with)
4342 The solution was to add an additional dependency on KFW
4343 in order or to be able to easily obtain the client principal
4344 name stored in the MSLSA ccache TGT. This information is
4345 used in two locations:
4347 - the pioctl() function
4349 - a new WinLogon Event Handler for the "logon" event.
4351 The pioctl function will now be able to use the correct
4352 name when calling WNetAddConnection2() and the "logon"
4353 event handler will now be able to call WNetAddConnection2().
4354 The hope is that the "logon" event handler will be called
4355 before the profile is loaded but I have not guarrantee
4359 * Fixed a locking error in cm_TryBulkStat() which had the
4360 potential of deadlocking the system for the length of time
4361 it takes to perform a bulk status fetch operation.
4363 * Replaced time conversion code (UnixTime <-> FILETIME) to
4364 be completely arithmetic instead of relying on a bizarre
4365 algorithm involving a variety of C RTL time functions.
4366 This has the side effect that UnixTime and FILETIME which
4367 are both stored in UTC are interpretted as UTC throughout
4368 the year. Windows will apply the same localization to AFS
4369 as it does to NTFS. Applications which rely on the ability
4370 to sync files between the two file systems will no longer
4371 see the timestamps of files in AFS change an hour relative
4372 to the files stored in NTFS or Windows based backup devices.
4374 * Fixed a invalid memory access under a bizarre circumstance.
4375 Windows will allow a physical mass media device to be
4376 installed via Plug N Play to the system and will assign it
4377 the lowest available drive letter. This is true even if the
4378 drive letter is currently assigned in the user session to
4379 a network device via NET USE (or its equivalent.) When this
4380 happens, queries sent to the CIFS server will contain invalid
4381 data. This invalid data was not being caught by the AFS
4382 Client Service and was resulting in a crash when Freelance
4385 * Fixed a reference count error when registering callbacks
4386 on a stat cache object if the callback was already registered
4388 * Add a case to cm_Analyze() in order to handle the case where
4389 cm_GetConnByMServers() returns CM_ERROR_NOSUCHVOLUME because
4390 the server list for the volume is empty. In this case, force
4391 an update of the volume info and retry.
4393 * Insert a missing cm_EndCallbackGrantingCall() which could
4394 result in threads waiting for a callback to complete to never
4397 * In the persistent cache, there is a maximum number of volume
4398 entries. Allow volume entries to be reused if the maximum
4399 number have been allocated and their reference count is zero.
4401 * If we already have a dead virtual connection object, don't
4402 ignore the fact that additional dead vc objects must be taken
4405 * Removed a deadlock condition introduced in the summer of 2004.
4406 Do not hold mx locks around calls to RX functions including
4407 rx_NewCall(). That is what reference counts on the rx_connection
4410 * Fixed an initialization error in afslogon.dll which could
4411 result in random behavior including a failure to terminate
4412 the NPLogon function. (1.3.80b)
4414 * Fixed an error preventing the use of SMB authentication
4415 on Windows 2000. (1.3.80a)
4418 * Updated CellServDB to Public CellServDB 16 Mar 2005
4420 * Fixed DNS AFSDB queries to ensure that "csail" cannot be
4421 misinterpretted as "csail.mit.edu" when the resolver
4422 is configured to append ".mit.edu" to failed DNS queries.
4424 * Fixed another case in which the client would replace connections
4425 to the host when it was not necessary. In this case the
4426 situation would occur if cryptall was on and the connections
4427 were unauthenticated due to lack of an appropriate token.
4429 * OpenAFS for Windows has failed to work at sites which are
4430 utilizing a cross-realm trust between an MIT/Heimdal realm
4431 and a multi-domain Windows forest when the workstation being
4432 accessed is not located in the root domain. This is caused
4433 by a bug in the workstation which was triggered after the
4434 introduction of Windows 2003 Server. When the bug is triggered,
4435 the workstation attempts to authenticate users to afsd_service.exe
4436 by contacting the Domain Controller instead of using the
4437 LSA loopback authentication mechanism.
4439 One of the reasons this bug occurs is because the workstation
4440 does not have a reliable way of knowing that the service whose
4441 netbios name is "AFS" is located on the workstation. This will
4442 be fixed starting in Longhorn Beta 1 by Microsoft. The
4443 "BackConnectionHostNames" registry value will be used to
4444 indicate that the authentications to that service name should
4445 be performed using the loopback authentication mechanism.
4447 In the meantime, when Logon Caching is enabled, we can force
4448 afsd_service.exe to authenticate using the logon cache before
4449 contacting the Domain Controller. This will work with both
4450 password and smart card based logons.
4452 * The allDown logic in cm_ConnByMServers() was wrong. The allDown
4453 flag should not be cleared if a volume's server reference is
4454 marked as "offline". In the case where all of the volume's
4455 servers are either "down" or the volumes are "offline", we want
4456 cm_Analyze() to process the condition as CM_ERROR_ALLOFFLINE
4457 instead of as CM_ERROR_TIMEDOUT. In fact, CM_ERROR_TIMEDOUT
4458 should never occur in practice.
4460 In the case of CM_ERROR_ALLOFFLINE, cm_Analyze() will sleep for
4461 5 seconds, clear the server down and volume busy flags, and
4462 then force an update from the VLDB. This allows the client to
4463 update the location of a volume if the reason for it being
4464 marked offline is because it is being moved. Calls to
4465 cm_ConnByMServers() will be retried either until success or
4466 the RDRtimeout period is reached.
4468 * Correct the Power Management code behavior in response to
4469 Standby, Suspend, and Shutdown events. Instead of flushing the
4470 buffers associated with the mounted SMB submounts, simply write
4471 all buffers which are marked dirty.
4473 * Added support for "TheseCells" to afslogon.dll. "TheseCells"
4474 provides a list of cells other than the default cell for which
4475 tokens should be obtained using the default Kerberos principal.
4476 This functionality is logon domain specific and is only available
4477 when using KFW for authentication.
4479 * Fixed FindNext Invalid Handle error caused by over agressive
4480 attempts at garbage collecting dirSearch entries when the
4481 dirSearch ID wraps from the maximum value to 1.
4483 * Add support for registry defined server preferences for VLDB and
4484 File servers. See registry.txt for details.
4486 * Increased default cache size to 96MB and default number of cache
4489 * Fixed refCount leaks related to directory lookups and pioctl calls.
4491 * Callbacks revoked during a race condition with an attempt to
4492 obtain the same callback no longer result in an inaccessible
4495 * New command line tool:
4497 afsdacl : Set or reset the DACL to allow starting or stopping
4498 the afsd service by any ordinary user.
4500 Usage : afsdacl [-set | -reset] [-show]
4501 -set : Sets the DACL
4502 -reset : Reset the DACL
4503 -show : Show current DACL (SDSF)
4505 * IP addresses are no longer queried once at startup. Instead IP
4506 addresses are obtained as needed. Loopback adapter addresses are
4507 no longer published to the server.
4509 * Pay attention to the MIT KFW registry configuration for automatic
4510 importation of MSLSA credentials. (SOFTWARE\MIT\Leash32,MsLsaImport)
4512 * Fix (once again) case-sensitive comparisons which was apparently
4515 * Activate support for vos listvol -format
4517 * Population of the "AFS Client Admins" group failed on non-English
4518 versions of Windows because the name of the "Administrators" group
4519 is localized. Now we lookup the name of the group by using the
4520 Administrators Group Alias Relative ID.
4522 * Multiple cell token acquisition within afscreds.exe was broken in the
4523 case of Kerberos 5 cross realm authentication. Instead of contacting
4524 the KDC associated with the cell's realm, afscreds would obtain a
4525 token from the KDC of the user's realm. This would result in
4526 "invalid kvno errors" while authenticating to the AFS servers.
4527 Unauthenticated access would work. The symptoms would vary based upon
4528 whether or not the VLDB servers had been contacted using unauthenticated
4529 connections prior to the user obtaining tokens.
4531 * The list of ACL entries was becoming corrupted. It appears as if
4532 an ACL when it expires was not being appropriately cleaned up. In
4533 fact, it was left in the list of ACLs associated with the scache
4536 * Changed the default @sys name list to "x86_win32 i386_w2k i386_nt40"
4537 for 32-bit x86 systems. The default for itanium will be "ia64_win64"
4538 and "amd64_win64" for amd 64-bit processors.
4540 * When executing executables, Windows provides a mechanisms by which
4541 the normal search for DLLs can be bypassed. If a file foo.exe is
4542 being executed and there exists a file or directory "foo.exe.local"
4543 (the contents are ignored if a file), then Windows will search for
4544 DLLs first in the "foo.exe.local" directory and second in the
4545 directory in which the "foo.exe" file is located.
4547 Previous releases of OAFW would improperly return
4548 STATUS_NOT_A_DIRECTORY instead of the expected
4549 STATUS_OBJECT_PATH_NOT_FOUND. This would cause the Windows SMB client
4550 to terminate the search for the DLL causing the execution of the
4551 application to fail.
4553 * Rework the reference counting of smb_vc_t objects. The references
4554 stored in the waiting locks were not counted. This could result in
4555 an assertion if the reference count drops to zero.
4557 * "fs wscell" returns "Freelance.Local.Root" for the workstation cell
4558 when running in freelance mode.
4560 * Added support for persistent caching of file contents, stat entries,
4561 volume data, cell data, and name lookup data. When the registry
4562 value, "NonPersistentCaching", is not defined or set to 0 the file
4563 specified by "CachePath" becomes a persistent cache file. The
4564 size of the cache file is computed at run time. It has a maximum
4565 size of approximately 1.2GB. The contents of the file will be
4566 validated according to the rules specified by the "ValidateCache"
4569 * Added support for UUIDs. UUIDs are kept for as long as the
4570 cache file is intact.
4572 * Added cmdebug.exe and support cache manager debugging callback
4573 interfaces in afsd_service.exe.
4575 Usage: cmdebug -servers <server machine> [-port <IP port>] [-long]
4576 [-addrs] [-cache] [-help]
4577 Where: -long print all info
4578 -addrs print only host interfaces
4579 -cache print only cache configuration
4581 * Symlinks to \\AFS\[all\]... will now be treated the same as
4582 symlinks to /afs/.... However, please use /afs/... as otherwise
4583 the symlinks won't work on Unix.
4585 * Correct a problem with local CellID allocation for cells whose
4586 server list is obtained via DNS instead of a CellServDB file. If
4587 the DNS information expires the CellID assigned to the entry will
4588 be changed. This causes all of the cm_scache_t objects which refer
4589 to the old cellID number to become useless. Attempts to access
4590 files or directories with cm_scache_t objects using the old cellID
4591 will fail since the server list cannot be obtained.
4593 * Correct deadlock condition in cm_EndCallbackGrantingCall which can
4594 be triggered if the volume referenced by the cm_scache_t object is
4597 * The AFS Service needs to respond to SERVICE_ACCEPT_SHUTDOWN messages
4598 in addition to SERVICE_ACCEPT_STOP.
4600 * Move RPC shutdown until after the SMB and RX shutdown procedures
4601 complete. Block until RPC shutdown is complete.
4603 * Modify afslogon.dll (integrated logon) to wait for service if its
4604 state is START_PENDING. If the timeout period occurs, reset to
4605 the retry interval and not the sleep interval.
4607 * When renewing the server list for a cell obtained via DNS AFSDB
4608 records, the cm_cell_t entry must be removed from the list of all
4609 cells. Otherwise, the list of cells will be corrupted.
4611 * In the dcache and scache modules, use the cm_scache_t dataVersion
4612 instead of the cm_fakeDirVersion.
4614 * Synchronize fs.c with the unix version.
4616 * The variable used to determine whether a file or virtual memory
4617 mapped cache is used was not properly initialized to a default
4618 value. If the registry setting "NonPersistentCaching" was not
4619 set, the choice would be random. Properly initialized to be
4622 * The memory mapped view was never unmapped before closing the file
4623 at service shutdown. This is now properly cleaned up.
4625 * Default location of Cache file is now %TEMP%\AFSCache
4629 * A bug affecting new installations of 1.3.75/76 would result in
4630 the creating of incorrect mountpoints in the freelance root.afs
4631 volume for the default cell.
4633 If "fs \\afs\all\<cellname>" lists a volume name of "root" instead
4634 of "root.cell", you have been affected by the bug. To correct the
4635 problem, execute the following commands:
4637 fs rmmount \\afs\all\<cellname>
4638 fs rmmount \\afs\all\.<cellname>
4640 fs mkmount \\afs\all\<cellname> root.cell <cellname>
4641 fs mkmount \\afs\all\.<cellname> root.cell <cellname> -rw
4645 * A bug has been fixed in the auto-generation of Freelance root.afs
4646 symlinks which produced random entries in the \\AFS\all directory.
4648 * Support has been added for multi-homed servers
4651 * Added a new registry value, "StoreAnsiFilenames", which can be used
4652 to force the use of ANSI character sets instead of OEM Code Pages.
4653 This feature is useful when users require the ability to create
4654 filenames with 8-bit characters and need to access the files from
4655 both Latin-1 based Unix systems as well as from Windows.
4657 Activation of this feature will prevent access to files stored with
4658 8-bit OEM characters.
4660 * Shutdown all SMB threads in a synchronized manner when stopping the
4663 * There is currently a maximum cache size of 1.3GB. The limit is imposed
4664 by the largest contiguous block of unused memory within the 2GB process
4665 space which can be assigned to the memory mapped file. Unfortunately,
4666 when the executable digital signature verification code is activated
4667 Windows sees fit to further segment the process memory which in turn
4668 reduces the size of the maximum cache file to less then 800MB. If
4669 larger cache sizes are desired, a new registry value should be set:
4671 HKLM\SOFTWARE\OpenAFS\Client (DWORD) "VerifyServiceSignature" = 0x0
4673 Setting this value will disable the runtime verification of digital
4674 signatures on afsd_service.exe and the afs dlls which it loads. It
4675 will not disable the the version number check on those same files.
4676 The signature verification is not a security messure and is only meant
4677 to enhance the ability to afsd_service.exe to detect potential
4678 destablizing mixtures of DLLs from incompatible distributions.
4680 Added code to auto-disable the signature verification check if
4681 the desired cache size is greater then 700MB.
4683 * Windows' WinTrustVerify(WIN_SPUB_ACTION_PUBLISHED_SOFTWARE) is
4684 used to verify the validity of the afsd_service.exe binary
4685 as well as each of the AFS DLLs loaded by the service. Not only
4686 must the digital signature be valid but the signatures of the
4687 DLL must be signed by the same entity as the service.
4689 * Implement new functions: cm_freelanceMountPointExists and
4690 cm_freelanceSymlinkExists. Use them along with other validity
4691 checks in cm_freelanceAddMount and cm_freelanceAddSymlink to
4692 ensure that name collisions do not occur and that empty strings
4693 are not valid file names.
4695 A symlink may not have a name which would resolve to a valid
4696 cell name. Doing so would prevent access to the cell.
4698 * Add missing cm_HoldSCacheNoLock call to Freelance mount point
4699 re-initialization code. The reference counts of the fake root.afs
4700 volume scache object(s) would become invalid when the mount point
4701 or symlink lists were altered.
4703 * Add registry entries to provide mappings from the afsdsbmt.ini
4704 to the new locations for applications which count on the use
4705 of the old Profile file APIs. These apps are likely to fail
4706 if the user does not have administrator privileges and the
4707 registry is locked down.
4709 * The afs_config.exe submounts dialog had two errors.
4710 First, attempts to remove entries failed because the registry
4711 key was being opened without KEY_WRITE privileges.
4712 Second, when editing a submount entry, changing the name
4713 would add a new key and leave the original one in place.
4714 Now the original submount will be removed if its name is
4717 * In recent months there have been several incidents in which
4718 users have experienced problems starting or accessing
4719 afsd_service.exe and after significant effort has been spent
4720 it has turned out that they have two versions of AFS on the
4721 machine or an inconsistent set of DLLs.
4723 Code has now been added to afsd_service.exe which will walk
4724 the list of modules loaded by afsd_service.exe and validate
4725 that the version of the AFS DLLs matches the version of the
4726 afsd_service.exe executable. If they do not match the service
4729 * When Freelance mode is enabled and there is no registry
4730 key HKLM\SOFTWARE\OpenAFS\Client\Freelance, afsd_service.exe
4731 will attempt to import the afs_freelance.ini file contents.
4732 If the file does not exist, it was creating a dummy file
4733 with a r/o and r/w entry for the default cell and then
4734 importing those values.
4736 This process has been changed. The temporary file is no
4737 longer created. Also, both the OpenAFS Client install
4738 directory as well as %WINDIR% are checked for previous
4739 afs_freelance.ini files.
4741 * Added support for VL_GetEntryByNameN(). Still need to add
4742 support for VL_GetEntryByNameU() for multi-homed support.
4744 * Fix a deadlock situation in afscreds.exe when canceling an
4745 auto-generated Obtain Tokens dialog
4748 * Fix the locking of objects during Directory Searches in the
4749 SMB/CIFS server. The failure to properly lock the reference
4750 counts was resulting in the premature freeing of smb_dirSearch_t
4751 objects while they were still in use by the SMB/CIFS client.
4752 This does not solve the "Invalid Handle" problem.
4754 * Fix Find Cell By Name pioctl call to return a valid cell
4755 name for the Freelance fake root.afs volume.
4756 "Freelance.Local.Root".
4758 * Fix the Explorer Shell Extension Symlinks->Add operation.
4759 The dialog template was missing and the link destination
4760 string was too short.
4762 * Add support for symlinks to Freelance root.afs volume
4763 Stored at HKLM\SOFTWARE\OpenAFS\Client\Freelance\Symlinks
4764 <number> = "<linkname>:<relative-path>."
4765 Use symlink.exe to create, list, or remove
4767 * Remove the fallback to the use of KFW's KRB4 library when obtaining
4768 tokens. We never obtain KRB4 tickets.
4770 * Fix AFS Client Configuration Control Panel to support new SysName
4773 * Fix a bug in afsd_service.exe which could result in the SysName
4774 not being read from the registry.
4777 * Add code to block the issuance of AFS tokens by aklog.exe or
4778 afscreds.exe when the Kerberos 5 principal name contains a dot.
4780 * Modify the IsAdmin() function to always treat the local SYSTEM
4781 account as an AFS client administrator. Affects fs.exe and
4784 * Modify the internal handling of Quota Exceeded errors
4786 * Upgrade all reference count fields in the Windows cache manager
4787 and the osi library to use unsigned long instead of signed short.
4788 A similar fix has been applied to the afs rpc (rx) library.
4790 * fix the Windows cache manager to prevent it from replacing the
4791 rx_connection object associated with the cm_conn_t object on each
4792 and every operation if "fs crypt" was set. This explains the
4793 dramatic performance difference when crypt is used vs clear.
4794 The problem: 'cryptall', a boolean flag indicating whether or not
4795 "fs crypt" is set, was being compared to the rx_connection
4796 cryptlevel which is either rxkad_clear:0 or rxkad_crypt:2.
4797 1 != 2 and therefore the rx_connection was always destroyed
4798 and replaced on each and every operation.
4800 Lock the cm_conn_t object around every call to RXAFS_xxxx functions.
4801 It is not safe for the cm_conn_t object to not be locked because
4802 rx_DestroyConnection might be called from another thread if:
4803 - the user's tokens have changed or expired
4804 - the crypt mode has changed
4806 This fix appears to have also taken care of the problems associated
4807 with Overlapped Writes resulting in Delayed Write errors.
4809 * fix NSIS installer's AdminGroup.exe to properly create and
4810 remove groups when given -create or -remove. The string comparison
4813 * fs sysname now accepts a list of sysname values
4815 * added a new registry value HKLM\SOFTWARE\OpenAFS\Client "IoctlDebug"
4816 DWORD which when set to a non-zero value will cause error message
4817 text to be output to stderr from the pioctl() routine. Useful in
4818 debugging failures of fs.exe, tokens.exe, etc.
4820 * added a test to the power management code to only perform a
4821 flush operation if there is at least one network adapter which
4822 is not a loopback adapter.
4824 * Fix bug in loading of registry value HKLM\SOFTWARE\OpenAFS\Client
4825 "EnableKFW". This value will not be read if the key
4826 HKCU\SOFTWARE\OpenAFS\Client exists; even if the "EnableKFW"
4827 value under that key does not.
4829 * provide mechanisms to force the use of krb524d for Kerberos 5
4830 ticket to AFS token conversion. For afslogon.dll and afscreds.exe
4831 there is a new registry value "Use524" and for aklog.exe a new
4832 command line parameter "-m".
4834 * Fix the pattern matching algorithm to properly match patterns
4837 * smb_ReceiveCoreRename() was factored to produce smb_Rename()
4838 which is used by both the original function and the new
4839 smb_ReceiveNTRename(). smb_ReceiveNTRename() supports the
4840 creation of HardLinks in addition to Renaming. smb_Link()
4841 is a new function which creates HardLinks via cm_Link().
4842 cm_Link() is a new vnodeops function which creates links
4845 smb_ReceiveNTRename() does not support the File Copy and
4846 Move Cluster Information operations described in its interface.
4847 ReceiveNTRename is under documented in CIFS-TR-1p00_FINAL.pdf.
4849 * When opening files via symlinks, we should follow the symlinks
4850 until we reach the actual file stat cache entry. The stat cache
4851 entry of the file should then be stored in the FID instead of
4852 stat scache entry of the symlink.
4854 * return bad operation errors for all unimplemented functions
4855 even if we do not know the functions exist.
4857 * Log bad packets and unknown operation packets to the trace log
4859 * Map CM_ERROR_BADOP to STATUS_NOT_SUPPORTED instead of
4862 * Update list of known CIFS operations to include all those listed
4863 in CIFS-TR-1p00_FINAL.pdf.
4865 * Modify the handling of HKLM\SOFTWARE\OpenAFS\Client\Submounts
4866 to support the REG_EXPAND_SZ type.
4869 * A new Windows authorization group "AFS Client Admins" is now
4870 created and populated with the members of the "Administrators"
4871 group. The group is used to determine which accounts on the
4872 machine may be used to modify the AFS Client Configuration via
4873 the UI and command line tools. afs_config.exe, fs.exe,
4875 * Modify the WinLogon Logoff Event Handler to query NT4 domain
4876 controllers for the remote profile path if Active Directory
4877 services are not available.
4879 * Fix aklog.exe to not add the AFS ID to the username
4881 * PTS registration of new users to foreign cells has been added to
4884 * The cm_Daemon thread is used to perform checks for
4885 down servers, up servers, volumes, callback expirations,
4886 lock maintenance and token expiration. Due to a gaff in
4887 larger integer division the thread never performed any
4888 work. Instead the current time computation would always
4889 be less then the trigger times. This had an adverse affect
4890 on the client's ability to maintain communication with servers,
4891 keep volumes up to date, and flush user tokens and acls
4892 when they have expired. This was broken when the 1.3 branch
4893 was modified to support VC7 which no longer included
4896 * An initialization problem with the Freelance code was
4897 detected while fixing the callbackRequest. The cm_rootSCachep
4898 object is obtained during afsd_InitDaemons() but the callback
4899 information is incomplete. The callback information will not
4900 be obtained until cm_MergeStatus is called from within
4901 cm_GetCallback. Unfortunately, cm_SyncOp did not properly
4902 test for the conditions under which the callback information
4905 * Reports have been filed indicating that callbacks were
4906 being lost. An examination of the code indicated that the
4907 cm_server_t objects were not being properly reference
4908 counted by the cm_scache_t and cm_callbackRequest_t objects.
4909 In particular, the cm_server_t objects may have been freed
4910 from beneath the cm_conn_t objects.
4912 All of the reference counting is now done via the functions:
4917 this improves the ability to track the referrals.
4919 Each cm_BeginCallbackGranting Call now allocates a reference
4920 to the cm_server_t. The cm_EndCallbackGrantingCall either
4921 frees the reference or transfers it to the cm_scache_t
4922 cbServerp field. These are then appropriately tracked
4923 through the cm_Analyze call.
4925 * Ensure that the dnlc hash table is the same size as the
4926 dir name hash table (as per original author's note).
4927 Increase the dnlc CM_AFSNCNAMESIZE to a multiple of 8
4928 for compatibility with 64-bit systems.
4930 * fix smb_ApplyV3DirListPatches to properly apply the hidden
4931 attribute to dotfiles when the infoLevel < 0x101 and
4932 cm_SyncOp has failed.
4934 * Fix the Freelance registry initialization code. There
4935 was a possibility that some systems could end up with
4936 garbage in the registry during a clean install.
4939 * file and directory names beginning with "." will now be given the
4940 hidden attribute when the volume access is anonymous. this matches
4941 the behavior when the volume access is via an authenticated user.
4943 * Added a change monitor to the HKLM\SOFTWARE\OpenAFS\Client\Freelance
4944 key. When a change occurs mark the root.afs data as invalid and
4945 for it to be reloaded on the next access. This allows administrators
4946 to modify the mount point list without restarting the service.
4948 The freelance client used to provide a fake modification time for
4949 the root.afs volume data and its mount points of 7/09/2001 14:24 EDT.
4950 Added code to extract the last modification time of the Freelance
4951 registry key and use that instead. The time now represents the
4952 most recent mount point change.
4954 * PTS registration of new users to foreign cells has been added to
4957 * Additional Cache Control and Credential Manager options have been
4958 added to the WiX installer. See deployment guide for details.
4960 * The CachePath setting is now optionally a REG_EXPAND_SZ type
4962 * The WiX installer has been upgraded. Version 2.0.1927.1 is now
4965 * The loopback installation code may have had a problem updating the
4966 %ETC%\HOSTS file which could have resulted in a premature failure.
4967 Work around code has been added for the case where the file cannot
4970 * The default max chunksize was increased from 15 (32K) to 17 (128K)
4971 because Windows sends 64K blocks when using overlapped writes.
4973 * The default number of server threads was increased from 4 to 25 to
4974 better handle overlapped writes.
4976 * The "AfscredsShortcutParams" registry value was not being properly
4977 loaded by afscreds.exe. Therefore, the default value was always being
4978 used instead of the value set by the installer.
4980 * Windows XP provides downgrade attack detection to prevent an attacker
4981 from being able to force the use of NTLM simply by disrupting
4982 communication with the KDC. This attack cannot exist between the
4983 Windows CIFS client and the AFS Client Service. Therefore, when a
4984 downgrade has been detected the afs pioctl library will force the
4985 establishment of a new CIFS connection using NTLM.
4987 * A locking error was discovered surrounding all references to volume
4988 server lists within the cm_cell.c source file.
4990 * The logged into Windows username was incorrect on Terminal Server
4993 * A new registry value "NonPersistentCaching" was added to the service
4994 parameters key. When set to a non-zero value, the afs cache is stored
4995 in the Windows paging file. There are two limitations to choosing
4997 1. when persistent caching is implemented it won't work with
4998 this flag set since there will be nothing to persist.
4999 2. with this flag set the initial paging allocation cannot be
5000 changed while the service is running
5002 * An initialization bug was discovered in aklog.exe which affected users
5003 who have a domain name for their afs servers which could not be mapped
5007 * afs_config.exe now validates cell names against DNS in addition
5008 to the CellServDB file.
5010 * In order to allow the freelance client to connect to a volume with ID
5011 equal to 1 on the default cell we changed the fake root.afs volume ID
5012 once again. This time we choose 0xFFFFFFFF. In addition, we change
5013 the cell ID of the fake root.afs volume from 1 to 0xFFFFFFFF as well.
5014 It will now be impossible for a volume ID to match that of another
5015 cell unless the client is connected to 0xFFFFFFFD cells. That should
5016 be enough room for growth.
5018 * Fix "fs mkmount" command to work with UNC paths and when
5019 started from non-AFS drives. It is now possible to create a mount
5020 point in the freelance fake root.afs volume with the command
5022 fs mkmount \\AFS\all\<directory-name> <volume-name> <cellname>
5026 fs mkmount \\AFS\all\openafs.org root.cell openafs.org
5027 fs mkmount \\AFS\all\.openafs.org root.cell openafs.org -rw
5029 * The algorithm used to re-attempt access to the servers associated with
5030 a volume has been altered to properly address the case in which all
5031 servers have been marked down. The previous algorithm did not reset
5032 the server's down flags so the servers were never actually retried.
5033 This caused a problem with active volumes if the network connectivity
5034 was lost as could be the case with a network cable removal, wireless
5035 drop, or laptop hibernation. With the fix volume access is restored
5036 almost instantenously when network connectivity becomes available.
5038 * Support for SMB/CIFS browsing has been added to the AFS Client Service
5039 SMB server. It is now possible to use "NET VIEW \\AFS" to obtain a
5040 listing of AFS submounts and freelance mount points. Support for
5041 NETSHAREENUM, NETSHAREGETINFO, NETSERVERENUM2, NETSERVERGETINFO
5042 significantly enhances the behavior of AFS volumes within the Explorer
5043 Shell. For instance, "AFS" now shows up as server in the Explorer
5044 with each submount or freelance mount point visible as a share.
5045 The right click menu in each folder now works with full functionality
5046 on a consistent basis.
5048 * The network provider can be configured to have different behavior
5049 depending on the domain that the user logs into. These settings are
5050 only relevant when using integrated login. A domain refers to an
5051 Active Directory (AD) domain, a trusted Kerberos (non-AD) realm or the
5052 local machine (i.e. local account logins). The domain name that is
5053 used for selecting the domain would be the domain that is passed into
5054 the NPLogonNotify function of the network provider. (see registry.txt
5057 * Added a new registry value [HKCU\SOFTWARE\OpenAFS\Client]
5058 "Authentication Cell" which may be used to specify a default
5059 authentication cell for afscreds.exe which is different from
5060 the default cell for the AFS Client Service daemon.
5062 * Added a Logoff WinLogon Event Notification function to afslogon.dll.
5063 afslogon.dll moved to %WINDIR%\System32\.
5064 New registry entries added to register the dll for Winlogon events.
5066 The logoff event will now force a call to ktc_ForgetAllTokens()
5067 using the context of the user being logged off as long as the
5068 user's profile is not loaded from within AFS. If the profile
5069 was loaded from AFS we can't release the tokens since the Logoff
5070 event is triggered prior to the profile being written back to
5071 the its source location. This is now performed in an XP SP2
5074 * Windows XP SP2 Internet Connection Firewall interoperability
5077 * The %WINDIR%\afsdsbmt.ini contains four sections:
5078 Submounts, Drive Mappings, Active Maps and CSC Policies.
5079 The Submounts and CSC policies are now stored in the registry under
5080 [HKLM\SOFTWARE\OpenAFS\Client\Submounts]
5081 [HKLM\SOFTWARE\OpenAFS\Client\CSCPolicy]
5082 The Drive Mappings and Active Maps are stored in the registry under
5083 [HKCU\SOFTWARE\OpenAFS\Client\Mappings]
5084 [HKCU\SOFTWARE\OpenAFS\Client\Active Maps]
5086 There is no automatic migration of this data as it would be impossible
5087 to consistently migrate data to user profiles which may not be active
5088 when the machine is updated.
5090 * The %WINDIR%\afs_freelance.ini contains lists of mountpoints for the
5091 fake root.afs volume. For the same reasons as for the cellservdb file,
5092 this information should not be in %WINDIR%. This information is now
5093 kept under the registry key
5094 [HKLM\SOFTWARE\OpenAFS\Client\Freelance]
5096 The data from the afs_freelance.ini file will be automatically
5097 migrated to the registry on first execution of afsd_service.exe
5099 * Keeping the CellServDB file in the location %WINDIR%\afsdcell.ini is
5100 troublesome for several reasons. One, it is confusing for those who
5101 expect the file to be named "CellServDB" instead of "afsdcell.ini".
5102 Two, this file is not a Windows Profile formatted file. Three,
5103 applications should not be reading or writing to %WINDIR%. It causes
5104 problems for Windows Terminal Server.
5106 The new location of CellServDB will be the OpenAFS Client install
5107 directory which is by default C:\Program Files\OpenAFS\Client and can
5108 be determined by querying the registry for
5109 [HKLM\SOFTWARE\TransarcCorporation\AFS Client\CurrentVersion]PathName
5111 The existing afsdcell.ini will be migrated by the NSIS installer.
5112 The Wix installer must still be updated to do the same.
5114 * Change NSIS installer to use DNS by default; to remove Integrated Logon
5115 High Security mode; and to add Terminal Services compatibility registry
5116 entries to allow the OpenAFS tools to find the afsdcell.ini and other
5117 configuration files in %WINDIR%.
5119 * Add support for authenticated SMB connections. This will remove
5120 the need for high security mode in most situations. Both NTLM
5121 and Extended Security (GSS SPNEGO) modes are supported. Effectively,
5122 only NTLM can be used even though Kerberos is now supported. The
5123 reason is that it is not possible to construct a service principal
5124 which is unique to each individual machine.
5126 SMB Extended Auth does not work on XP SP2 unless one of two registry
5127 modifications are made:
5129 (1) To disable the check for matching host names on loopback connections
5130 set this key. This does not require a reboot:
5132 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa]
5133 "DisableLoopbackCheck"=dword:00000001
5135 (2) To add the AFS SMB/CIFS service name to an approved list. This
5136 does require a reboot:
5138 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa\MSV1_0]
5139 "BackConnectionHostNames"=multi-sz "AFS" "MACHINE-AFS"
5141 afsd_service.exe will automatically add the current Netbios Name
5142 to the BackConnectionHostNames list and then temporarily disable
5143 the loopback check for one cycle of startup/shutdown of the service.
5144 We assume most folks do not start/stop without a reboot so this
5145 will be adequate in most cases.
5147 * Fix security hole in afslogon.dll which allowed passwords to be
5148 sent in clear text to the KDC in a misformed principal name.
5150 * Fix cm_GetCell() to properly handle expired dns entries
5153 * If Freelance mode is active and the afs_freelance.ini
5154 file does not exist, do not create an empty file.
5155 Instead create a file containing ro and rw mountpoints
5156 to the default cell using the standard conventions.
5158 * Modify the Freelance support to handle the ability
5159 to create rw mount points in the fake root.afs volume.
5161 * Changed the RPC mechanism used for token setting from
5162 named pipes to local. Use of named pipes can be restored
5163 by setting the environment variable AFS_RPC_PROTSEQ to
5166 Named pipes were required when a Windows 9x system was
5167 using a NT system in gateway mode which is incompatible
5168 with our use of local loopback adapters.
5170 * In afscreds.exe, if a username of the form user@REALM is
5171 specified and no password is specified, do not perform a
5172 kinit operation. Only perform the aklog functionality.
5174 * Add a new registry value which allows the number of processors
5175 on which afsd_service.exe executes to be restricted. Valid
5176 values are 1..numOfProcessors
5178 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
5182 * A second MSI based installer option is now available.
5184 * Fixed Kerberos 5 kinit functionality in afscreds.exe to properly
5185 request tickets for user/instance@REALM instead of just user@REALM
5187 * Modify the Power Management Notify routine to wait for the Hard Dead
5188 timeout period instead of a fixed 19 seconds. With the longer timeout
5189 periods Hibernation and Standby could never succeed when network
5190 connectivity is not available.
5192 * The following fs.exe commands are now restricted to Administrator:
5193 - checkservers with a non-zero timer value
5196 - sysname with a new sysname list
5205 setting the default sysname for a machine should be done via the
5206 registry and not via "fs sysname".
5208 * NSIS installer adds options to install Debugging Symbols
5209 and the Microsoft Loopback Adapter; the user is now also
5210 given the ability to select the afscreds.exe startup options.
5212 * Build system modified to generate symbols for FREE (aka RELEASE)
5213 builds as well as CHECKED (aka DEBUG) builds
5215 * Sites which have a volume ID of 0x20000001 assigned to their
5216 root.cell volumes have been experiencing problems with accessing
5217 the root.cell volume of their cell when Freelance mode has been
5218 active. This was because 0x20000001 was assigned to the fake
5219 root.afs volume created by freelance. The fake volume id is
5220 now set to 0x00000001 to prevent conflicts.
5222 * The timeout logic in the AFS Client Service has been wrong
5223 for sometime. It is based on two different assumptions.
5224 First, the SMB client timeout is a fix value as was the case
5225 with OS/2 Lan Manager. This assumption is incorrect. The
5226 SMB timeout in Windows is a dynamic value computed based upon
5227 a fixed minimum timeout to which is added time based upon the
5228 size of the request and the performance characteristics of
5229 the connection. Second, it is the responsibility of the
5230 SMB Server to enforce the timeout requirements of the client.
5231 This is untrue. The SMB Server cannot be expected to know
5232 the requirements of the client. More importantly, if the
5233 SMB server uses the SMB client timeout as a value to restrict
5234 its behavior as an RX client, the performance characteristics
5235 of the local SMB session would be used to prematurely terminate
5236 WAN connections with significantly different performance
5239 The timeout logic has therefore been modified in the following
5241 . the Lan Manager Workstation (SMB) Session Timeout is used only
5242 as a basis for configuring the Connection Dead Timeout
5243 and Hard Dead Timeout values. The Connection Dead Timeout
5244 must be at least 15 seconds longer than the SMB Timeout
5245 and the Hard Dead Timeout must be at least double the
5246 Connection Dead Timeout.
5247 . New registry entries have been added to allow the Connection
5248 Dead Timeout and Hard Dead Timeout values independent of the
5249 Lan Manager Workstation Session Timeout
5250 . The test to enforce the SMB Client Timeout has been removed.
5252 One of the side-effects of removing the enforcement of the SMB
5253 Client Timeout is that regardless of whether or not the SMB client
5254 is available to receive the response (and how would the SMB server
5255 know) the RX protocol response can be used to update the AFS
5256 Client Service state for ready access by future SMB client
5259 This should be the end of the "Server paused or restarting messages"
5261 * Add support for arbitrary UNC paths to the pioctl() support.
5262 This enables the fs commands as well as the AFS Shell Extension
5263 to work correctly when UNC paths are being used.
5265 * Fix afscreds.exe (by updating afskfw.lib) to search for cells via
5266 DNS if the cell configuration cannot be determined via CellServDB
5268 * Add debug info to test whether CM_BUF_WAITING or CM_SCACHE_WAITING
5269 are ever set more than once at a time
5271 * Fix the management of lists of cm_cell_t structures when using
5272 DNS to lookup cell information. The previous code would fail to
5273 reuse the same cellID for a cell if DNS was used more than once
5274 for a given cell name. When the ttl expired, a single cm_cell_t
5275 could be inserted into the cm_allCellsp list more than once
5276 producing a loop. In addition, the vlServerp list belonging to
5277 the cell was not freed resulting in improper refCounting of the
5280 * Add DNS support to cm_IoctlNewCell() which previous only examined
5283 * Add cm_FreeServer() function and call it from cm_FreeServerList()
5284 to properly garbage collect cm_server_t objects
5286 * Add numVCs variable to smb.c to track the number of smb_vc_t
5287 objects created and use it to initialize the vcID field which
5288 previously was set to 0 in all objects resulting in FindByID
5291 * Fixed DNS lookups to work consistently throughout the OpenAFS
5292 product instead of just from within the afsd_service.exe
5294 * Added a runtime check to ensure that AFS Client Service SMB
5295 Server is accepting connections before attempting to mount
5298 * Read IP addresses for volume servers out of the CellServDB
5299 file if gethostbyname() on the hostname fails.
5301 * Fix getcellconfig() to populate both the Hostnames as well
5302 as IP addresses when loading cell data via DNS
5304 * Increase the Connection Dead Time to 50 from 20 seconds
5305 Increase the Hard Dead Time to 120 from 40 seconds
5306 (matches the Unix values)
5308 * Fixed an assertion validating the number of allocated NCBs
5310 * Fixed the build environment to consistently build for
5311 Windows 2000 and above. (APPVER = 5.0)
5313 * Fixed rx_debug to properly validate the receipt in incoming
5314 data with select() and recvfrom(). Do not copy data out of
5315 the socket buffer unless success is indicated.
5318 * afsd_service.exe will now display a message box to the
5319 desktop when it terminates due to an IP Address Change.
5321 * installer no longer deletes AFS Server configuration data
5324 * installer generates a warning dialog if the RPC service
5325 is not properly configured
5327 * installer compressed with lzma instead of bzip2
5329 * afsd_service.exe shutdown crash solved once and for all
5331 * reference counting of smb_vc_t data structures improved
5333 * name space collision of smb_fid_t event objects corrected
5335 * the output of "fs memdump" is now written to
5336 %WINDIR%\TEMP\afsd_alloc.log
5338 * the file TaAfsApp_1033.dll is now properly installed allowing
5339 the User Manager to start
5341 * a new algorithm is used for computing filename pattern matches
5343 * afscreds.exe now accepts user names containing instance
5346 * Fix the Directory Name Lookup Cache to be case-sensitive.
5347 This is crucial in environments in which a Windows client
5348 is accessing a directory with more than one filename that
5349 differs only by case. If the directory contains "FOO"
5350 and "Foo". You want "DEL Foo" to delete the correct one.
5351 We still have a problem in that "DEL foo" will delete a
5352 random filename. This will be addressed in a future release.
5354 * Fix afscreds.exe -M option (renewMaps) to work when High
5355 Security mode is off. Also, remember to disable the ActiveMap
5356 flag in afsdsbmt.ini when a drive mapping is removed.
5358 * Updates to NSIS installer script. AFS Server configuration
5359 data will not be destroyed on un-install or re-install.
5360 Use a better compression algorithm.
5362 * afslogon.dll now uses KFW to obtain tokens when available
5364 * afslogon.dll when given an all uppercase username will
5365 attempt to authenticate with both the uppercase name
5366 and an all lowercase variation
5368 * DST modification removed. The fix appears to make things
5369 worse after a reboot of the machine.
5371 * fs.exe: added "cscpolicy" which is used to
5372 change client side caching policy for AFS shares
5374 Usage: fs cscpolicy [-share <AFS share>] [-manual] [-programs]
5375 [-documents] [-disable] [-help]
5377 * Several uninitialized variables have been initialized
5379 * It is now possible to obtain tokens using cross realm
5380 Kerberos within afscreds.exe:
5382 user: jaltman@ATHENA.MIT.EDU
5384 Will obtain a cross realm ticket for jaltman/DEMENTIA.ORG@ATHENA.MIT.EDU
5385 will will in turn be used to obtain afs@DEMENTIA.ORG.
5386 The resulting token will be stored with the display name
5387 jaltman@ATHENA.MIT.EDU@dementia.org
5389 * aklog.exe has been added to the client
5391 Usage: aklog [-d] [[-cell | -c] cell [-k krb_realm]]
5392 [[-p | -path] pathname]
5396 -d gives debugging information.
5397 krb_realm is the kerberos realm of a cell.
5398 pathname is the name of a directory to which you wish to authenticate.
5399 -noprdb means don't try to determine AFS ID.
5400 -5 or -4 selects whether to use Kerberos V or Kerberos IV.
5401 (default is Kerberos V)
5402 No commandline arguments means authenticate to the local cell.
5405 * All of the resource files have been restructured to adhere to
5406 a set of rules IBM implemented for loading string resources.
5407 These rules had either been forgotten or were not discovered
5408 by folks working on the OpenAFS sources. The end result was
5409 memory corruption. This is primary item which was preventing
5410 the AFS Server from working.
5412 * Increased the size of the maximum ticket size stored in a token
5413 from 344 bytes to 12,000. Increased the buffers used to convey
5414 messages between the pioctl() caller and the SMB Server from
5415 1000 bytes to 12,512. The code appeared to have been writing
5416 above the top of the stack by quite a few number of bytes.
5417 (The increased ticket size is necessary for the next item.)
5419 * When obtaining AFS Tokens via KFW, krb524 is no longer required.
5420 Instead the raw Kerberos 5 ticket is used in its entirety. This
5421 is extremely important as it allows us to use pure Kerberos 5 KDCs
5422 as the source of the AFS authentication. The use of up to 12,000 byte
5423 tickets will allow tickets produced by all versions of Microsoft
5424 Active Directory to be used.
5425 - create a user account.
5426 - designate it DES only
5428 - specify its UPN to be "afs@realm"
5429 - assign a SPN of "afs/cellname" to the UPN with setspn.exe
5431 * Do not enforce the funky 8dot3 pattern matching rule that the first "."
5432 is special when using long file names. (you must use "*.*" and not "*")
5433 Instead only enforce it when performing 8dot3 searches.
5435 * Fixed the DST problem with creation times being set one hour ahead
5437 * Fixed the problem when using \\afs\cell-alias. For example,
5438 \\afs\uncc instead of \\afs\uncc.edu. Do not a new cell struct
5439 for the alias name; instead simply expand the name. One of the
5440 symptoms of this problem was a loss of acquired tokens.
5442 * Fixed the AFS Shell Extension. The Symbolic Link menu was empty
5443 of strings. (Only English strings provided.)
5445 * Fixed the installer to properly replace in use files.
5447 * Fixed the build system to cleanup generated component version files
5449 * The release build compiled with MSVC 6.0 compiler to avoid the
5450 afsd_service.exe shutdown crash. This does not solve the problem
5451 but simply avoids it for the time being.
5455 * fix afslogon.dll to not corrupt memory when High Security mode
5458 * fix afsd_service.exe to not attempt to restore the stack when
5459 an exception occurs. (not safe in multi-threaded programs)
5461 * fix uninstaller to properly remove the CRT and MFC DLLs
5463 * remove a Message Box from afscreds.exe when getcellconfig()
5464 fails on a kerberos realm which is not a cell
5466 The following is a list of changes to the OpenAFS for Window client
5469 * "fs setserverprefs" will leave afsd service deadlocked
5471 * "vos listaddrs" will core dump
5473 * installer sets the appropriate keys to support Integrated Logon
5475 * installer disables the "Find Lana by Name" functionality as it
5476 was causing headaches for many users
5478 * fix the intermittent crash of the power management thread when
5479 shutting down the AFS Client Service
5481 * optimizes the obtain drive mount list functionality which is
5482 executed every time the mount tab in afscreds.exe and afs_config.exe
5483 are refreshed. (this happens a lot)
5485 * fix the service shutdown logic. add the STOP_PENDING state
5486 and do not accept additional service events after we declare
5489 The following is a list of changes to the OpenAFS for Window client
5492 * flexelint was run against the source tree and hundreds (perhaps
5493 thousands) of corrections were applied to ensure prototypes
5494 were in use; types were used consistently; variables were
5495 initialized; unused variables were removed; etc.
5497 * A wide variety of instrumentation was added including the
5498 ability to produce a stack trace from within afsd_service.exe
5501 * Dynamic configuration of the RDRtimeout value based upon the
5502 LanMan Workstation Session Timeout
5504 * The mount root no longer needs to be called "/afs". This
5505 is now set by a registry value "MountRoot" within the key
5506 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5508 * The cell list is now only read out of afsdcell.ini when the
5509 file changes instead of each time a cell is resolved.
5511 * Thread synchronization was added to cm_server.c and ktc_nt.c
5513 * All calls to GlobalAlloc()/GlobalFree() were replaced with
5514 calloc()/free(). The Global functions were needed on Windows 3.x
5515 but have caused a variety of problems on the Win32 platforms.
5516 Avoiding them is highly recommended by several Microsoft
5517 Knowledgebase articles
5519 * Support for Symbolic Links added to the AFS Shell Extension
5521 * Added a registry value "OverlayEnabled" to determine if
5522 Shell Extension Overlays should be enabled.
5523 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5525 * New Build system to support VC6, VC.NET, VC.NET2003 compilers and
5526 separate trees for checked and free builds. Build system supports
5527 a custom directory src\WINNT\extra which can be used as a grafting
5528 location of organization specific additions to the build tree.
5530 * New installer built using NSIS 2.0.
5532 * Named all kernel objects in order to allow them to be monitored
5533 with tools such as SysInternals' ProcExp.exe.
5535 * Introduced new EventLog framework for AFSD
5537 * Introduced Power Management interface to AFSD for Standby and
5538 Hibernate modes to allow cache to be flushed prior to network
5541 * Utilize Win32 DNSQuery API instead of internal routines. This
5542 allows DNS SRV queries to be sent to all current domain name
5543 servers. Not just one specified in an INI file. DNS is now
5546 * "NetbiosName" registry value may be used to specify a fixed
5547 Netbios Name such as "AFS" to be used instead of "HOSTNAME-AFS"
5548 when the loopback adapter is in use. If you need to use the
5549 old notation with a loopback adapter installed specify a registry
5552 "NetbiosName" REG_EXPAND_SZ = "%COMPUTERNAME%-AFS"
5554 * Refactor all modules which depend on LAN Adapter and NetbiosName
5555 determination in a new library: lanahelper.lib. This allows for
5556 consistent behavior throughout the product.
5558 * Move the afsd.log and afsd_init.log files to the directory specified
5559 by the "TEMP" environment variable. This is usually %WINDIR%\TEMP
5560 for services. Added the Date to the log entries.
5562 * New registry value "RxMaxMTU" used to limit the size of the RX
5563 packets sent by the AFS Client Service to the Server. In order
5564 to enable OpenAFS to work across the Cisco IPSec VPN the packet
5565 size must be restricted to 1264 or smaller. The latest NSIS
5566 installer sets a value of 1260 by default.
5568 * New registry value "RxNoJumbo" to disable the use of Jumbo Rx
5569 packets. This is not needed in order to work across the Cisco
5570 VPN but might be needed for other network environments. This
5571 value is not set by the NSIS installer.
5573 * New registry value "HideDotFiles" is used to apply the Hidden
5574 attribute to files whose names begin with a '.'. This value
5575 is set by the NSIS installer.
5577 * New registry value "MaxMpxRequests" allows the maximum number
5578 of multiplexed sessions to be configured at run time. This
5579 value is not set by the NSIS installer. The default value is
5582 * New registry value "MaxVCPerServer" allows the maxmimum number
5583 of VCs per server to be configured at run time. This value is
5584 not set by the NSIS installer. The default value is 100.
5586 * New registry value "AllSubmount" allows the "all" submount to
5587 be disabled by setting its value to 0x00.
5589 * Allow cells names to be valid mount points
5590 \\<netbiosName>\<cellname>
5592 * Store the active state of drive mappings in order for afscreds.exe
5593 to restore them upon startup
5595 * Add exception handling to generate a Stack Trace to the afsd_init.log
5596 file if one happens to occur.
5598 * Add lots of logging to help detect the cause of invalid SMB packets
5600 * Enable Kerberos for Windows to be used to obtain AFS Tokens via
5601 conversion of Kerberos 5 "afs" service tickets. Supports auto-
5602 renewal of expiring tokens as long as afscreds.exe is running.
5604 * New afscreds.exe command line options:
5606 -M = renew drive maps
5607 -N = ip address change detection
5610 * New registry value "EnableKFW" in {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5611 determines whether or not MIT Kerberos for Windows should be used
5612 to obtain tokens via Kerberos 5 tickets.
5614 * New registry value "AfscredsShortcutParams" in
5615 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5616 determines the command line parameters to be specified when "fixing"
5617 the AFS Shortcut in the user's startup folder.
5619 * The "ShowTrayIcon" registry value has been moved from
5620 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5621 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5623 * The <cell name> registry values used to store the token expiration
5624 reminders have been moved from
5625 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5626 {HKCU,HKLM}SOFTWARE\OpenAFS\Client\Reminders
5628 * Obtain the Logon User Name from the Explorer key when available
5630 * new text document doc\txt\winnotes\registry.txt lists all registry
5631 values used by OpenAFS (excluding the AFS Server)
5633 * BUG: rx_securityClass objects were not properly reference
5634 counted and were never freed.
5636 * BUG: reduce the number of conditions under which CM_ERROR_TIMEOUT
5637 would be generated. The existence of a server does not imply
5638 that it is not down. If all of the servers for a cell are down
5639 return CM_ERROR_NOSUCHVOLUME instead. This prevents the Explorer
5642 * BUG: the directory name lookup cache failed to free the entries
5643 in the cache when the name cache entries cycled. The entries
5644 in the cache would become dereferenced without being freed.
5646 * BUG: fs setserverprefs could be executed without Administrator
5649 * BUG: the number of allocated NCB objects (100) exceeded the number
5650 which could actually be waited upon by the kernel (64). Any objects
5651 which were utilized above the limit could never have event completions
5654 * BUG: smb_username_t objects were not being reference counted and
5655 were not properly freed.
5657 * BUG: smb_tid_t objects could under unusual circumstances be freed
5658 before they were no longer referenced.
5660 * BUG: smb_fid_t object pointer were frequently used even when
5661 their value could be NULL. They were not properly released and
5662 therefore they were never freed.
5664 * BUG: smb_packet_t data structures were not completely initialized
5667 * BUG: when Rx produces a CM_ERROR_NOIPC error do not return "Access
5668 Denied" because that causes the Explorer Shell to try again until
5669 access is obtained. Instead return "Remote Resources" which allows
5670 the shell to move on and treat the error as transient.
5672 * BUG: when initializing the NCBreturns structure, separate Event objects
5673 were created for each NCB although a single Event object was supposed
5674 to be shared by all.
5676 * BUG: smb_dirSearch_t objects were not being properly referenced counted
5679 * BUG: smb_tran2Packet_t objects were not being properly referenced
5682 * BUG: directory path creation did not handle the case of multiple
5683 directories requiring creation in one attempt
5685 * BUG: SMB requests which required an Extended Response were ignored.
5686 This prevented some files from being written to AFS volumes.
5688 * BUG: character strings were being freed even after they were
5689 inserted into in use data structures
5691 * BUG: inconsistent usernames were used when High Security mode was
5692 enabled. (there is still much to do in this area)
5694 * BUG: pioctl() calls which require out of band RPC operations were
5695 susceptible to race conditions when performed by multiple processes
5697 * BUG: memory allocation and deallocation crossed instances of the
5698 C Runtime Library producing memory leakage and corruption in
5699 afscreds and the client configurator.