1 Since 1.5.9906 (1.6pre6)
2 * add support for detection of NTFS symlinks to
3 \\AFS using GetFileInformationByHandleEx which is
4 available on Vista, Server 2008 and beyond.
6 * improve thread safety of afs_shl_ext.dll
8 * MS11-043 adds response validation for SMB_COM_NEGOTIATE
9 messages received by the SMB Redirector. OpenAFS failed
10 to properly specify a Challenge and DomainName in the
11 response when the security mode is SMB_AUTH_NONE (or share
12 with password). This patchset corrects smb_ReceiveNegotiate()
13 so that it adheres to the protocol specification.
15 * smb_T2SearchDirSingle() must not fail directory search requests
16 for the _._AFS_IOCTL_._ file. Although this file does not actually
17 exist, it is successfully processed by CreateFile operations.
18 Therefore, an explicit search for it should return a valid answer.
20 * cm_SyncOp/cm_SyncOpDone is used to synchronize the RPC processing
21 to ensure that calls which are in conflict cannot occur at the
22 same time but also to ensure that the ordering of operations
23 is consistent. cm_MergeStatus() was in many cases executed after
24 cm_SyncOpDone() removed the synchronization barrier which in turn
25 permitted status information to be applied out of order. Side
26 effects could have included data loss due to client side file
27 truncation. More commonly two StoreData RPCs would have their
28 status information applied out of order forcing the cache manager
29 to invalidate all of the cached data for the file.
31 * cm_NewServer() can result in a call to cm_UpdateVolumeLocation()
32 if a server probe is performed. In order to avoid recursive
33 calls to cm_UpdateVolumeLocation() do not probe new servers from
34 within cm_UpdateVolumeLocation().
36 * Numerous Rx improvements.
38 * Fix out of order lock acquisition in smb_ReceiveV3LockingX()
39 which occurs if the lock acquisition fails.
41 Since 1.5.9905 (1.6pre5)
42 * aklog supports dotted Kerberos v5 principal names.
44 * afskfw library always attempts afs/cell@USER-REALM
46 * afskfw library must test return code from
47 krb5_cc_start_seq_get() or will trigger a null
48 pointer exception when using Heimdal.
50 * lock protected fields must be 32-bit in order
51 to avoid memory overwrite races.
53 Since 1.5.9904 (1.6pre4)
54 * Fix caching of non-existent volumes. The test to
55 trigger an immediate CM_ERROR_NOSUCHVOLUME in
56 cm_UpdateVolumeLocation() was backwards.
58 * Prevent the background daemon from checking the
59 status of non-existent volumes. cm_CheckOfflineVolumes()
60 should skip volume groups with the CM_VOLUMEFLAG_NOEXIST
63 * The afskfw library should return an error immediately
64 if the krb5_32.dll library cannot be loaded. Affects
65 afslogon.dll and afscreds.exe.
67 * No longer depend on leashw32.dll in afskfw library.
69 * NPLogonNotify must provide the user password in all
70 calls to KFW_AFS_get_cred(). It cannot count on a
71 credential cache being preserved between calls. Permits
72 tokens to be acquired for all cells listed in the
73 TheseCells registry value for a domain.
75 * Improve the trace logging from NPLogonNotify().
77 * Avoid a race when writing the cm_scache_t mountPointString
78 when acquiring mount point or symlink target data via
79 cm_GetData(). The race could result in bogus target
82 * Permit the use of des-cbc-md5 and des-cbc-md4 enctypes
83 as DES keys in asetkey.exe.
85 Since 1.5.9903 (1.6pre3)
86 * Fix automatic addition of Freelance cell mount points
89 * Avoid recursive offline volume checks which could cause
90 a crash due to stack exhaustion.
92 Since 1.5.9902 (1.6pre2)
93 * rx calls issued on a busy call channel are now specially
94 handled and logged. All calls are automatically retried.
96 * The vos support for multi-homed servers was incorrectly
97 implemented. The changes have been reverted.
99 * Avoid wasting cm_buf_t objects and associated data buffers
100 for obtaining mountpoint and symlink target strings. Instead
101 read them directly into the cm_scache_t mountPointTarget
104 * Optimize the evaluation of mountpoint and symlink targets
105 by skipping the FetchStatus and a full round-trip if
106 the cache manager does not have valid status information
109 * Fix the abstraction of cm_FreelanceAddMount() by passing
110 in volume names that do not include the trailing dot.
113 [there was no 1.6pre1 for Windows]
115 * vos commands now manipulate servers by UUID and can
116 recognize multi-homed servers.
118 * afs_config will not longer set the Tray Icon State
119 in the registry if the checkbox is not present in
120 the dialog. [RT 128591]
122 * Heimdal's roken utility library has been added
123 as \Program Files\Common\afsroken.dll
125 * When probing servers to determine if they are up or down
126 no longer issue an RXAFS_GetTime RPC to servers that indicated
127 that they do not understand the RXAFS_GetCapabilities RPC.
128 Since they responded it is known that they are up.
130 * AFS Explorer Shell Extension now works from folder
131 backgrounds. Overlays for mount points and symlinks
132 are present in the dll, but are not registered at present
135 * Do not use RankServerInterval registry value as the value for
136 PerformanceTuningInterval.
138 * CellServDB updated to 13 Dec 2010 release from grand.central.org
140 * Add "fs chmod" command and display current mode as part
141 of "fs examine" output.
143 * When the data version of a mountpoint or symlink changes,
144 the target string in the cm_scache_t object must be cleared.
145 Otherwise, the new target will not be queried.
147 * "fs checkservers" now includes vldb servers in the output
148 and only lists multi-homed servers once. A multi-homed
149 server that has at least one up interface is no longer
150 considered to be down.
152 * When asynchronously storing dirty data buffers to the
153 file server ensure that (a) the cm_scache_t object and
154 the cm_buf_t object are for the same File ID so that
155 locking and signalling work properly; and (b) if the
156 FID no longer exists on the file server, do not panic,
157 just discard the buffer.
159 * When processing VNOVOL, VMOVED and VOFFLINE errors perform
160 server comparisons by UUID or address and not simply by
161 cm_server_t pointer. Otherwise, server failover may not
164 * Do not preserve status information for cm_scache_t objects
165 when the issuing server is multi-homed.
167 * Giving up all callbacks when shutting down or suspending
168 the machine is now significantly faster due to the use
169 of an rx_multi implementation. (This functionality is
170 still off by default and must be activated by a registry
173 * Race conditions were possible when updating the state
174 of the cm_volume_t flags and when moving the volumes
175 within the least recently used list.
177 * Ensure that the lanahelper library does not perform a
178 NCBRESET of each lan adapter when enumerating the
179 current network bindings. Correcting this permits OpenAFS
180 to work on Windows 7 when the network adapter settings
183 * Fix creation of mount points and symlinks as \\AFS\xxxx
187 * Reimplement FetchData and StoreData operations to make
188 use of rx_Writev and rx_Readv which is faster than than
189 multiple calls to rx_Write and rx_Read. Throughput
190 improvements of up to 20% are possible.
192 * Many improvements to the rx rpc stack:
196 - Fast recovery is not triggered by packet resend timeouts
198 - Socket errors are now detected and reported
200 - avoid repetitive memory allocations
202 - avoid calling gettimeofday() for every packet
204 - reduced dependency on the call->lock
206 - When a call receive is completed, ack all of the packets
207 immediately. Do not wait for the first response packet
210 - Do not hold the call->lock during rx_Write* and rx_Read*
211 operations when the call fields in use can only be accessed
212 by the application thread.
214 - Reduce the default window size from 128 back to 32 because
215 the current overhead from walking packet queues in the rx
216 listener thread significantly hampers the ability to
217 process incoming packets from the network.
219 - Only backoff the peer timeout once for a series of naks.
221 - Release builds are lean and mean. All debugging and
222 packet tracking code is disabled.
224 * Version number is logged to the Windows Application Event
225 Log as part of the start pending message.
227 * MSI installers now backup and restore configuration
228 data during upgrades.
230 * Rx pmtu discovery has been disabled by default. Use
231 HKLM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
232 RxPMTUDiscovery "DWORD"
233 value to re-activate.
235 * cm_BkgDaemon event evaluation has been optimized to avoid
236 unnecessary overhead.
238 * If a volume is known to be readonly, then locally fail any
239 change request without sending it to the file server. This
240 avoids triggering the file server's abort threshold.
242 * Do not leak cm_volume_t objects from the volume LRU queue.
244 * Ensure that cm_NameI errors are acted upon promptly.
245 There are many cases in the SMB server where an error from cm_NameI()
246 was either ignored or not acted upon until several other operations
247 are performed that could result in the same error being repeated.
248 This is a mistake which did not have negative side effects until
249 additional checks for callback status were added recently.
251 At present, if a CM_ERROR_ACCESS error is returned and ignored,
252 subsequent attempts to operate on the same cm_scache_t will result
253 in additional queries to the file server that will also end in an
254 abort response. This can trigger the file server to delay responses
257 * Fix fs mount point and symlink create operations which were
258 failing due to improper directory separator termination.
260 * RXAFS_GetVolumeStatus can return VNOVOL, VMOVED, etc. In
261 order to process them and update volume state a fid must
262 be passed to cm_Analyze(). Use the volume root fid.
264 * Negative caching for volume lookups that fail with either
265 VL_NOENT or VL_BADNAME. The expiration time is five minutes.
266 This prevents volume lookup storms.
268 * Provide each volume lookup its own two hour lifetime instead
269 of reseting all volume location information every two hours.
270 This prevents the first smb path evaluation after the volume
271 location data has been discarded from potentially causing an
275 * When the SYSTEM account is in use, the SMB connection is
276 not always established as S-1-5-18. Sometimes it uses an
277 anonymous, S-1-5-7, connection. In both cases the SMB
278 authentication name is the nul string. To better identify
279 when an SMB connection is local system, the smb_username_t
280 is now stored as a SID instead of a user name. When a SID
281 is used, a new flag bit is set SMB_USERNAMEFLAG_SID.
283 In smb_SetToken if AFS_PIOCTL_AFSLOGON is set and the user
284 connection is not SYSTEM, the RPC SID is checked. If RPC
285 SID is SYSTEM, that supercedes the SMB authentication.
287 * Replace STATUS_IO_TIMEOUT with RPC_NT_SERVER_TOO_BUSY for
290 * Replace STATUS_PATH_NOT_FOUND with RPC_NT_SERVER_UNAVAILABLE
291 for CM_ERROR_ALLOFFLINE and CM_ERROR_ALLBUSY.
293 * Optimize behavior of buf_CleanAsyncLocked(). Avoid searching
294 for the cm_scache_t object by FID if it is already known.
296 * If the readonly file attribute is set (stored as a unix mode)
297 then a CreateFile operation should fail if the file is opened
298 for DELETE in combination with any other privilege.
300 * If the directory buffer contents are garbage we can crash
301 the service. Add some simple validation checks to ensure
302 that cm_dirEntry_t objects have the correct flag value and
303 that the name strings are not too long.
305 * If the bulkStat errorCode indicates that a particular object
306 is inaccessible due to a VIO error, we must update the server
307 status appropriately in order to permit failover.
309 * An RX_MSGSIZE error is returned by the new PMTU detection
310 code. It is critical that such an error result in a retry of
311 the operation that failed. Otherwise, the PMTU detection can't
312 work and the server will be marked down.
313 Secondly, it is important that such errors not leak to the
314 application layer. Map them to CM_ERROR_RETRY in all cases.
316 * Translate RX_RESTARTING to the same cache manager error code
319 * When logging server volume instance errors to the windows
320 application event log, be sure to log the cell as well.
321 Translating from server ip address is non-trivial. Make it
322 easier for administrators triaging issues to plug the volume
323 and cell info into vos commands.
325 * Since unix mode bits are represented in octal in most cases
326 make sure we log them that way.
328 * The cm_enforceTrailingDot() function failed if there was a
329 trailing dot followed by white space.
331 * Rx: Do not send RX_MSGSIZE if the PMTU size did not change.
334 * A crash could occur if a single letter share name was
335 dynamically evaluated as matching a cell name. This was
336 due to a failure to treat the comparison of two nul
337 strings as identical.
339 * Docbook validation is performed by xmllint during builds.
341 * Volume package bugs in the file server can result in VOFFLINE
342 being returned to the client instead of VNOVOL or VMOVED. As
343 a result the Unix CM treats VOFFLINE the same as VMOVED and VNOVOL.
344 The Windows client has not. As a result, bugs in the file server
345 can cause the Windows client to lose if the volume has in fact
346 been moved to another server.
348 As part of this change, the volume location list is updated prior
349 to the volume status being applied to the server from which the
352 * If the pages to be flushed are from a readonly or backup volume
353 they can't be dirty. In this case there is no need to stabilize
354 the pages before they are locked and recycled.
355 Stablilization is performed on the cm_scache_t object so do not
356 stabilize and unstablize for each cm_buf_t object.
358 * Over the years the processing of the Freelance callbacks have
359 added functionality that behaves much more like FetchStatus checks
360 to a file server. If the data version of the object has changed,
361 get the new data. Given that is the case, we can remove much of
362 the original refresh logic that is rather race prone. Say goodbye
363 to cm_fakeGettingCallback and cm_fakeDirCallback.
365 * When processing a pioctl path with either smb_ParseIoctlPath or
366 smb_ParseIoctlParent, cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK|GETSTATUS)
367 must be called on the cm_scache_t object to ensure that it is up
368 to date before we permit cm_Lookup or other operations to be performed
369 on it. Add the cm_SyncOp() call to smb_ParseIoctlPath and
370 smb_ParseIoctlParent to ensure it is done for all pioctl operations.
372 * The 32-bit tools installer was displaying client configuration wizard
373 pages. Since, the 32-bit tools share configuration with the AFS
374 service, the 32-bit tools installer should not be attempting to change
375 client configuration. Remove said wizard pages from the 32-bit tools
378 * cm_data.rootSCachep is a global pointer to the cm_scache_t that
379 represents the root.afs volume root directory. Throughout the
380 code this cm_scache_t was being used without ensuring that a
381 callback to the volume is in fact valid.
383 Under most circumstances this would not be a problem. However,
384 it is possible for a request to fail due to the lack of a callback
385 at a critical moment. Add a new function cm_rootScachep() that
386 attempts to ensure that a callback is present (if possible) prior
387 to use the cm_scache_t object.
389 * When the contents of the Freelance root directory changes the fake
390 directory buffers are updated and a fakeDirVersion is incremented.
391 The dataVersion of the cm_scache_t object is supposed to be updated
392 on the next access by performing a fake get callback request.
393 Unfortunately, this did not always occur because of a race. If another
394 Freelance object is updated first, the root directory object would
395 never successfully get a fake callback.
397 This patchset ensures that the generation of the fake directory
398 buffer content and the callback are obtained under the same set
399 of locks thereby removing the race.
401 * Enforce in afsd_service that tokens can be set for alternate
402 userids only if the SetToken pioctl is called from the
403 Local System account.
406 * Revise SMB QuerySecurityInfo Response for MS10-020
408 MS10-020 (http://support.microsoft.com/kb/980232) has caused
409 many problems for implementors of SMB 1.0 servers and applications
410 that call GetFileSecurity() without checking the return code to
411 determine if the call succeeded. The gist of the vulnerability
412 was that the SMB redirector would pass any buffer it received
413 to the application regardless of whether or not it was valid.
414 MS10-020 protects the applications by strictly validating the
415 SMB response data structure and the data in the security descriptor
418 The problem for SMB 1.0 server implementors is that there have
419 been at least three different protocol descriptions for
420 NT_TRANSACT_QUERY_SECURITY_DESC published over the last decade
421 and all of them are incomplete. Therefore, just about no one but
422 Microsoft has an SMB 1.0 server implementation that produces the
423 exact out that they are expecting to validate.
425 The end result is that in an attempt to protect applications from
426 crashing due to invalid input being passed in directly caused
427 dozens of applications to crash by not returning any security
428 descriptor data at all. Even when the applications didn't crash
429 they might not have been able to save their data. Cisco WAAS
430 and NetApp DataOnTap systems were most adversely affected and
431 they have had CIFS protocol licenses for many many years.
433 To fix OpenAFS here is what needed to be done:
435 1. Instead of returning a security descriptor that gives ownership
436 to the NUL SID, give it to the Everyone SID and set the flag
437 that states that everyone has full access.
439 2. Validate the input parameters. In particular, check to ensure
440 that the SMB file descriptor is valid and the file has not
443 3. Enforce the maximum output data and parameter counts.
445 4. Handle buffer overflow and buffertoosmall conditions
446 in the manner that Microsoft expects them to be handled.
447 In particular, note that the parameter data which is returned
448 in the SMB Data Region is not counted in the Data Count.
449 Even if MaxData is 0, we can still return parameters values
450 as long as MaxParm is large enough.
452 * Prevent use of AFSCache file contents if mapped to
455 * The Windows version of "fs newcell" did not accept any parameters
456 and behaved quite differently from the Unix version. Instead of
457 permitting new cell information to be added, the Windows version
458 simply forced the existing cell information to be reacquired.
460 This update adds a new pioctl, VIOCNEWCELL2, to support the
461 implementation of a Unix-style "fs newcell". The functionality
462 added here differs from the Unix version in the following ways:
464 1. "fs newcell" with no arguments is still accepted
465 in order to maintain compatibility with prior Windows
468 2. "fs newcell -cell <cell> -dns" instructs the cache manager
469 to add the new cell but obtain the vldb server info from
472 3. "fs newcell -cell <cell> ... -registry" instructs the cache
473 manager to add the new cell and also save the cell configuration
474 data in the registry for use the next time the service restarts.
476 4. The -vlport and -fsport options are accepted although the
477 -fsport value is currently unsupported by the cache manager.
479 * New registry value "FreelanceImportCellServDB" instructs Freelance
480 to create a mount point for every cell name listed within the
483 * Path MTU discovery for Rx is activated.
485 * Rx socket input buffer is converted to a circular buffer.
487 * Fix usage of cm_FreeServerList(). Do not set the server list
488 pointer to NULL after calling cm_FreeServerList(). Doing so
489 can result in a memory leak.
491 * Only enable Rx NAT pings on a single anonymous connection at a
494 * Fix cm_IoctlSkipQueryOptions() buffer management. Prevents a
495 potential read beyond end of memory buffer.
497 * Reduce requested privileges when reading registry CellServDB
498 to the minimum required.
500 * Add support for RPC Pipe Service NetWkstaGetInfo levels
501 101 and 102 which are called on Windows 7 and 2008-R2.
503 * Prevent integer overflow during quota percent used calculation
504 in Explorer Shell Extension (RT 126846)
506 * Generate a meaningful error if "fs listacls" or "fs setacls"
507 are executed on the Freelance root.afs volume.
509 * RXAFS_InlineBulkStat errors must be processed via cm_Analyze.
510 RXAFS_InlineBulkStatus does not return errors such as EACCES,
511 VNOVOL, VNOVNODE, VOFFLINE, VBUSY, VIO, VMOVED, etc. as an RPC return
512 code. Instead they are returned in the status info errorCode field
515 Traditionally, the error associated with the first FID in the query
516 list has been returned to the caller of cm_TryBulkStatRPC().
517 However, the error has never been processed through cm_Analyze()
518 which means that the per-vnode processing for VNOVNODE and the volume
519 global processing for VMOVED, VNOVOL, etc. has never been performed.
520 As a result, failover to other .readonly volume instances cannot occur,
521 volume moves will not be handled, and files that have been deleted
524 This patchset makes the following changes:
526 1. If an inline bulk operation has been performed and the inline
527 errorCode is a volume global error, then that error replaces
528 the RPC return code within the cm_Analyze() processing for
529 the RPC. This will affect whether or not a retry operation
532 2. The variable 'inlinebulk' is reset to 0 at the top of the
533 cm_Analyze() loop in case failover from an inlinebulk capable
534 file to an inlinebulk incapable file server takes place.
536 3. The FID that is passed into cm_Analyze() is not a real fid.
537 Instead it consists of the cell and volume but vnode = 0.
538 This ensures that the error (if any) is not applied to the
541 4. If an inline bulk operation was performed, prior to performing
542 the cm_MergeStatus() operation a vnode a check is made to
543 determine if an error was returned for that vnode. If so,
544 cm_Analyze() is called with no connection, a fake cm_req_t,
545 the fid, and the error. This permits cm_Analyze() processing
546 to be performed on the file.
548 * Show configuration pages for all types of MSI installations
550 The OpenAFS MSI installer wizard used to not show any configuration
551 pages for "Typical" and "Complete" installations. Setting the
552 workstation cell and logon options during installation required
553 selecting the "Custom" option. Many users choose the "Typical" option
554 during installation, and thus would never see the configuration pages.
555 Therefore, for these users, the workstation cell was being set to the
558 This patch makes the workstation cell and logon option configuration
559 pages visible to all types of installations (except silent
560 installations which show no UI).
562 * cm_LookupInternal creates Freelance mount points and symlinks
563 when queries cannot be found in the Freelance root.afs directory.
564 If the search name is a full cell name for which vldb information
565 can be obtained, then a mount point is added. If the search name
566 is a left-most substring or the full cell name with a dot appended
567 to it, then a symlink was created. This approach created a very
568 poluted Freelance name space.
570 This patchset makes the following changes:
572 1. Do not create symlinks with a dot appended to the cellname
574 2. Do not create symlinks where the left-most substring is not
575 a full dot separated component of the cellname.
577 3. Permit lookups to succeed when we would have created a
578 symlink in the past without creating the symlink.
580 * BPlus tree lookups are much faster than searching through
581 the native directory format on Windows because the case sensitive
582 hash tables cannot be used successfully. Permit BPlus trees
583 to be used except when called with cm_BPlusDirFoo as the action
584 function because cm_BPlusDirFoo is used to build the BPlus trees
585 from the native directory format.
587 * Symlinks are ending up in the Freelance root.afs directory that
588 end with a dot. Make sure it cannot happen.
590 * cm_FreelanceAddMount and cm_FreelanceAddSymlink is supposed to
591 return the allocated FID of the entry that was added. However,
592 cm_NameI is called to perform the lookup without forcing an update
593 of the Freelance fake directory. As a result the entry may not be
596 Force an update prior to calling cm_NameI() by using
597 cm_clearLocalMountPointChange() and cm_reInitLocalMountPoints()
600 * The Freelance fake root directory buffers were not zero-filled.
601 This results in random behavior that can cause the service to
602 terminate unexpectedly.
604 * The validation check for the response from the GetVolumeStatus
605 pioctl is incorrect. The response is not simply a VolumeStatus
606 structure but also several C strings appended to it.
608 * When flushing a file, we need to commit the file length changes
609 as well as the dirty buffers. Call cm_FSync instead of buf_CleanVnode
610 which is called by cm_FSync.
612 * Prevent rx_rpc_stats global lock from being a bottleneck in the
617 * Avoid a race when updating cell vldb server lists
618 that can result in a crash.
620 * Avoid a deadlock when managing CM_SCACHESYNC_STOREDATA
621 state operations for directory objects.
623 * Add new Windows Application Event log messages for
624 VBUSY, VRESTARTING, ALL_BUSY, ALL_OFFLINE, and ALL_DOWN.
625 Include message throttling to prevent the same message
626 from being logged repeatedly within a five second window.
628 * Reduce lock contention by waiting for cm_buf_t I/O
629 operations to complete before permitting cm_SetupStoreBIOD
630 to analyze a buffer for inclusion in a BIOD.
632 * Split the cm_buf_t flags field to separate the flags
633 that are protected by the cm_buf_t mutex from those
634 protected by the buf_globalLock. This eliminates the need
635 to hold both locks everytime the flags field is accessed.
636 Both locks were not held in the past resulting in race
637 conditions that could result in deadlocks.
639 * Add "vos setaddrs" command.
641 * Rx library lock contention avoidance between rx_NewCall and
644 * Rx library races due to inconsistent use of rx_connection
645 conn_data_lock to protect the flags field.
647 * Rx library inconsistent use of RX_CALL_TQ_WAIT which could
650 * Rx library must signal transmit queue waiters when flushing.
651 Otherwise, deadlocks can occur.
653 * In cm_UpdateVolumeLocation, avoid searching for a ".readonly"
654 volume on a numeric volume name.
656 * File buffer allocations whose offsets are beyond server EOF
657 should be locally allocated and zero filled. The file server
658 should not be issued a FetchData rpc which is guaranteed to
661 * Enable integrated logon to work with Windows 7/2008 when
662 user logons are performed with a non-Domain Kerberos principal.
664 * Add Protection Error messages to aklog output.
668 * Prevent the Explorer Shell extension from crashing if
669 symlink creation failed. (126406)
671 * A Rx level NAT ping has been implemented.
672 Add NatPingInterval registry value to
673 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
674 to permit Nat Ping to be enabled. The default value is 0 seconds.
676 * When a re-initialization is taking place, be sure to reset
677 cm_noLocalMountPoints to 0 in case someone deletes the "Freelance"
678 registry key out from underneath the service.
680 * Add krb5 error message translation to aklog, afscreds,
681 afslogon.dll, the network identity manager afs provider
684 * Mode bits aren't directly exposed by the Win32 API. We were leaving
685 them to default to 0777 when creating new files and directories.
686 This version introduces two configuration parameters;
687 'UnixModeFileDefault' and 'UnixModeDirDefault' which are DWORD
688 registry entries that are used to set the initial mode bits.
689 If the values are set to 0, then the behavior is identical to what we
692 * Minidump files are now produced with a timestamp appended
695 * An SMB request debugging monitor has been added. When activated
696 the monitor will automatically turn on trace logging if any SMB
697 request has required longer than 60 seconds to complete and will
698 then create a minidump every 60 seconds thereafter until the
703 * Restore use of DNS AFSDB and SRV records by kaserver clients.
707 * Avoid a potential Freelance deadlock during initial execution
708 of afsd_service.exe if the old ini file data has to be
711 * Three rx library corrections. (1) Idle data connection
712 processing could timeout if the send window filled and
713 took longer than the idle data timeout period for the
714 transmit window to re-open. (2) The transmit queue
715 could be emptied prematurely. A required check for the
716 queue being in use was forgotten. (3) The function that
717 is supposed to implement a wait for the transmit queue
718 to cease being busy failed to wait.
721 * Restore use of DNS AFSDB and SRV records which were
722 unintentionally disabled in 1.5.69
725 * Add a context menu to the NetIdMgr AFS Provider
728 * Prevent an empty directory Btree from being created and
729 marked as valid if cm_BPlusDirBuildTree fails.
731 * cm_BPlusEnumAlloc should not fail if the enumeration
732 contains zero entries.
734 * In cm_BPlusXXX functions, return ENOMEM if malloc() fails.
736 * Do not leak the cm_scache_t dirlock if cm_BPlusDirBuildTree
739 * fs examine should report owner and group ids as signed values.
741 * Protect buffers in smb_WriteData from simultaneous writes.
743 * Query network interface MTU values from the registry because
744 the IP Helper GetAdapterAddresses() fails to report a manually
747 * Improve performance by not dropping and reacquiring the rx_call
748 lock in rx_WriteProc* and rx_ReadProc*.
750 * When configuring the rx library, the network interface MTU
751 must be applied to both the maximum send size as well as the
752 maximum receive size as is done when rx_SetMaxMTU is called
755 There was no 1.5.67 release
758 * Convert afsauthent.dll to use the xdr_alloc and xdr_free
759 routines from afsrpc.dll in place of malloc and free.
760 This permits the two libraries to be built with different
761 C runtime libraries without crashing.
763 * Set the DOS Readonly attribute on a file/directory
764 whenever the unix mode combined with the mask 0200
765 is true. Previously there was a discrepency between
766 the mask used for testing for readonly behavior and
767 that used for setting the attribute.
769 * Disable AFSVolSync based .readonly "whole-volume callback"
770 support because the all file servers prior to 1.5.67
771 (and perhaps 1.4.12) do not properly assign a value to
772 the AFSVolSync structure in bulk status RPC responses.
774 * Improve the error output from aklog to output the value
775 from krb5 error_message() if the afs_com_err output
776 indicates an unknown value.
778 * Hold a lock on the cm_scache_t object when the smb
779 server is assigning a new clientModTime value.
781 * Permit custom build version numbers to be set
782 without modifying the src/config/NTMakefile.<platform>
785 * Add support for binary code signing with cross-signed
786 certificates and arbitrary versions of signtool.exe.
788 * Convert VBUSY and VRESTARTING to CM_ERROR_ALLBUSY and
789 do not permit them to be exposed to the smb redirector.
791 * Convert STATUS_TIMEOUT responses to STATUS_IO_TIMEOUT
792 to avoid confusion within the smb redirector.
794 * Fix the byte order assigned to port numbers associated
795 with AFSDB record lookups. They must be network byte
796 order not host byte order.
798 * Add dynamic server ranking based on RPC round trip
802 * Official support for Windows 7 and Server 2008 R2
804 * Improved Rx RTT computation following the lessons
807 * Prevent a file server bug (FetchData returning an
808 invalid length instead of zero) from causing an
809 "unexpected network error" when writing to files.
811 * Promote DNS SRV records as superior to DNS AFSDB
812 records. Support arbitrary port numbers for vldb
815 * Add "fs listacl -cmd" support.
817 * Add AFSVolSync based .readonly "whole-volume callback"
818 support. With this functionality, multiple objects from
819 a .readonly volume can have their status validated by
820 issuing a single RXAFS_FetchStatus RPC.
822 * Remove drive mapping functionality and service start/stop
825 * Remove drive mapping functionality from afs_config.exe.
827 * Adjust SMB error return codes to avoid returning
828 STATUS_TIMEOUT which results in the SMB redirector
831 * Network Identity Manager OpenAFS Provider now provides its
832 own "AFS lock" notification icon to report the status of
833 "have tokens, have no tokens, service not started, service
834 started but inaccessible". Hovering over the icon lists the
835 cells for which tokens exist (if any) and the OpenAFS version
836 number. Double-clicking executes the Network Identity Manager
839 * When merging the current status for an AFS object, ensure
840 that the volume status for that object is set to 'vl_online'.
842 * Permit cm_Analyze to request a retry even when the cm_req_t is
843 flagged CM_REQ_NORETRY if the RPC failed due to RX_CALL_DEAD.
844 The retry will force the use of a new connection.
846 * Add digital signatures to resource only dlls as required by
849 * Prevent pioctl calls from retrying indefinitely when a sharing
850 violation error occurs.
852 * Add {HKLM,HKCU}\SOFTWARE\OpenAFS\Client DWORD "ShowMountTab"
853 to restore access to drive mapping functionality in afscreds.exe
856 * Permit Rx minimum peer timeout, maximum receive window size and
857 maximum send window size to be configured via the registry.
859 * The MSI installer now sets the ARPINSTALLLOCATION as required by
862 * DNS lookups set the port numbers in network byte order within
865 * Replace use of the 16-bit compatibility API WinExec with
866 ShellExecuteEx in afs_config.exe and afscreds.exe. WinExec
867 is incompatible with Windows 7 / Server 2008 R2 UAC
871 * Short circuit background volume checks if a shutdown
872 or suspend request is received.
874 * Avoid contacting the file server in order to allocate
875 an empty buffer beyond the current length of the file.
877 * When a request for streams on a directory or mount
878 point object is received, do not offer a default stream.
879 This was affecting the saving of roaming profiles.
882 * Fix Windows 2000 compatibility. Do not call rand_s()
883 even if it is compatible with the C run time library
886 * Fix a data consistency error between the output of
887 NetWkstaGetInfo and NetServerGetInfo RPCs, specify the Lan
888 workstation group name "AFS", and report server name as
889 "AFS" instead of "\\AFS" when the caller asks for "\\AFS".
891 * Enable executables to be run from \\AFS on Windows 7.
892 Return "Name not found" instead of "File not found" when
893 a directory or file name cannot be found.
895 * Prevent cache manager from marking file server "down" when
896 the data returned in response to either RXAFS_FetchData64
897 or RXAFS_StoreData64 is invalid.
899 * Add pioctl data validation to the AFS Explorer Shell extension.
902 * After 1.5.62 afslogon.dll would no longer crash but it
903 also would not load "domain" specific configurations.
906 * An uninitialized variable was present in the symlink
907 recursion detection loop.
909 * Reverse addition of unique per file GUID during
910 Create/Open responses. Inclusion of the GUID prevents
911 Cygwin applications (and others) from accessing //afs.
913 * Treat "filename::$DATA" as a synonym for "filename".
914 This addresses the recent report of roaming profiles
915 failing to save back to the file server.
917 * When given a choice between an error from rx_Write()
918 or RXAFS_EndStoreData(), use the rx_Write() error.
919 This ensures that over quota and access denied errors
922 * If an error occurs during RXAFS_StoreData operations,
923 apply that error to all of the buffers in the BIOD.
924 This ensures that failures will not be retried if they
925 will only fail again.
927 * Do not perform a DNS lookup for a cellname if the
928 cellname prefix is _._AFS_IOCTL_._. This will avoid
929 DNS lookups when the query is for
930 _._AFS_IOCTL_._.F7E5F580200909010061TTTTNT7TT.{10E39A49-4531-4496-A08E-842D4C440D20}.
932 * Fix Freelance root.afs volume object callback processing.
933 Object synchronization and status merging bugs.
935 * Fix Explorer Shell "invalid parameter" bug introduced
938 * Log RXKAD errors to the trace log.
940 * Prevent simultaneous pioctl calls from multiple processes
941 from stepping on each other's data.
943 * Prevent simultaneous pioctl calls from crashing the
946 * Add data validation to pioctl output processing in
947 ktc_GetToken and ktc_ListTokens.
949 * Add data validation to pioctl output processing in
950 fs.exe and symlink.exe.
953 * IMPORTANT: New support for DCE RPC Services: SRVSVC and WKSSVC.
954 No longer will browsing \\AFS produce truncated share names.
956 * Registry specified server preferences did not result in
957 the CM_SERVERFLAG_PREF_SET flag being applied to the
960 * Avoid unnecessary DNS lookups of share names as cells
961 when it is known that the name cannot be a cell name.
962 Any name that does not contain a dot is skipped.
964 * When processing Dfs Referral requests, do not return an
965 error if the server is down or busy; if the volume is
966 offline; the cell vldb cannot be reached; etc.
967 These paths are still \\AFS paths and so the client
968 should not be sent to search elsewhere for resolution.
970 * CRITICAL: Prevent the cm_Daemon thread from terminating when
971 the machine enters the suspend state. IF the cm_Daemon thread
972 dies, there is nothing to execute down server checks.
974 * Prevent use of smb_StartedLock before initialization which
975 can result in a panic in the lock package.
977 * Improve error checking in afslogon.dll in order to prevent
978 NULL pointer dereferences within GetLogonDomainOptions()
979 if the Lsa operations fail.
981 * Add support to the build system to automatically update
982 a Microsoft Symbol Store.
984 * Do not access the cm_conn_t after executing the cm_Analyze
985 loops. Doing so can result in an access to memory that has
988 * CRITICAL: If a StoreData request offset is not aligned on the
989 buffer module blockSize, (offset % blockSize) bytes of dirty
990 data will not be written to the file server.
992 * CRITICAL: If a StoreData64 request is sent to a file server
993 that does not support large files (>2GB) and the client is
994 unaware that the server has no large file support, the first
995 chunksize worth of data will not be written to the file server.
996 This bug was introduced in 1.5.3. It affects all IBM AFS file
997 servers and all OpenAFS file servers older than 1.4.0.
1000 * If a file server becomes inaccessible while the cache manager has
1001 dirty buffers to write, the afsd_service buf_IncrSync thread can
1002 attempt to use 100% of the cpu.
1004 * Fix the locking that protects the cm_cell_t lists
1005 (all, name, id, free).
1007 * Remove cm_cell_t objects from the name and id hash tables before
1008 placing them on the free list.
1010 * Fix "fs newcell" which was broken in 1.5.60.
1012 * Do not attempt to synchronize dirty buffers if the associated
1013 volume is known to be unavailable.
1015 * Log to the Windows Event Log dirty buffers that cannot be written
1016 during the shutdown process.
1018 * Modify behavior of a Freelance mountpoint target that does not
1019 specify a cell. Instead of assuming the target volume is in the
1020 Freelance.Local cell, use the workstation "Cell" specified in the
1021 registry. A mountpoint target of "#root.cell." will now mean the
1022 root.cell volume in the workstation cell for the current session.
1023 If the workstation cell changes from "athena.mit.edu" to
1024 "andrew.cmu.edu", the referenced volume will also change without
1025 requiring that the mount point targets be altered.
1027 * Avoid false infinite loop errors when validating the cm_cell_t
1028 free list during startup.
1030 * Fix two more locations where xdr_free() should be used instead of
1031 free() to permit a checked built of afsd_service.exe to work with
1032 release builds of afsrpc.dll.
1034 * Add cm_FindServerByUuid(). Re-implement RXAFS_InitCallBackState3()
1035 to permit the server Uuid to be used to lookup the server object
1036 and from that determine the cell. This permits callbacks that are
1037 received from alternate addresses to be processed with a known server
1038 object. Previously a request from an unknown server would clear all
1039 callbacks from all cells.
1041 * Fix a bug that prevented optimal performance when using a non-zero
1042 value for 'daemonCheckVolCBInterval'. As a reminder, when
1043 "daemonCheckVolCBInterval" is set to a non-zero value, all .readonly
1044 volume callbacks are automatically renewed 90 minutes before their
1047 * Fix automatic ranking of vldb servers whose values are obtained from
1048 the CellServDB file.
1050 * Add logging for RX CALL DEAD errors which are generated if the remote
1051 host is unreachable.
1053 * Add failover for RX CALL TIMEOUT errors when the volume is readonly
1054 or the call is to a vldb server.
1056 * Add registry based cell search functionality to NetIdMgr,
1057 afs_config.exe, and klog.exe.
1059 * afsconf_GetCellInfo() has been modified to perform gethostbyname()
1060 lookups on the host names in the CellServDB instead of using the
1061 specified IP addresses. This provides aklog, pts, vos, etc. the same
1062 CellServDB behavior that the Windows Cache Manager uses.
1064 * Completely a behind the scenes change. The snprintf() routines have
1065 been completely rewritten.
1067 * When updating the stat cache entry callback of a .readonly object
1068 from the volume group object, update the file server reference to
1069 ensure it matches the most update to date callback.
1071 * Add proper support for processing callbacks from multi-homed file
1072 servers. Instead of comparing servers by cm_server_t pointer,
1073 compare them by UUID when the UUID is known.
1075 * During a shutdown short circuit the offline volume check daemon
1078 * Properly track link counts during cm_Link and cm_Unlink operations.
1080 * Return the error code of RXAFS_FetchData / RXAFS_StoreData in
1081 preference to an error code reported by rx_EndCall.
1083 * Add "PerFileAccessCheck" registry value to permit testing against
1084 experimental file servers that include per-file acl support. This
1085 value is intentionally undocumented. It is not to be used by
1086 production environment deployments.
1088 * Log a warning to trace log if a volume group is marked as a DFS
1089 File Set. The Windows OpenAFS client does not include DFS support.
1091 * Fix a bug introduced in 1.5.60 that prevents the afs netidmgr
1092 provider from obtaining tokens when referrals are in play.
1094 * Add "fs chown" and "fs chgrp" commands to permit the owner and
1095 group of objects stored in AFS to be set from Windows.
1097 * Avoid performing background daemon operations when the machine is
1098 going into suspend mode.
1100 * Perform offline volume checks in most recently used order.
1102 * Prevent crash when a data version for a cache object goes backwards.
1104 * fs setquota, fs setcachesize, vos setfields, and vos create now
1105 accept human readable orders of magnitude. (K, M, G)
1107 * fs listquota fixed to permit large quota sizes to be displayed.
1109 * rx packet resend and data packets sent counts were incorrect.
1111 * Multi-thread safe library versions are now being generated and used.
1112 mtafsubik.lib, mtafsutil.lib, mtafsvldb.lib, mtafsvol.lib.
1114 * Microsoft SMB Redirector (mrxsmb.sys) support for ExtendedSessTimeout
1115 values are now available on XP through Windows 7. Add functionality
1116 to autodetect if such support is present on the machine. If so,
1117 configure it if necessary and dynamically adjust the AFS Rx timeout
1120 * Correct another race condition in the Rx library that could result
1121 in an unexpected panic while freeing the Rx call iovq.
1124 * A fix to the pioctl library to support drive substitution
1125 to UNC paths. (SUBST <d:> <\\afs\cell\path>).
1127 * On April 9th Microsoft released a Hot Fix for Windows Server 2003 SP2
1128 that corrects a deadlock in the smb redirector and also adds new
1129 functionality that permits the AFS SMB server to be given a longer
1130 timeout than is normally the case. New functionality has been added
1131 to configure these additional LanmanWorkstation\Parameter values.
1132 (This functionality has been backported to XP SP3 and is scheduled
1133 to be released on June 5th.)
1135 * The BackConnectionHostNames registry value configuration was broken
1136 when dynamic re-establishment of Netbios Name registrations was added.
1139 * Hidden vos commands are revealed.
1141 * If the "DisableLoopbackCheck" registry value is set, do not unset
1142 it during the same service session.
1144 * Reorganize code that prevents multiple Store operations to the same
1147 * Modify IsPathInAfs test in Explorer Shell Extension and fs.exe to
1148 permit broken symlinks to be treated as being in AFS.
1150 * Fix vos commands that output 64-bit integer values.
1152 * Cygwin Import Libraries are provided in the SDK for all OpenAFS DLLs.
1153 Permits building cygwin applications against OpenAFS libraries.
1155 * OpenAFS Release Notes, Administrator Guide and User Guide now installed
1156 as Windows HTML Help (.CHM) files.
1158 * NSIS installer does a much better job of cleaning up files left
1159 over from previous installs.
1161 * Fix RT#124787, a race condition between "fs flush <dir>", "fs flushvolume",
1162 or "fs flushall" and on-going directory operations that can result in
1163 afsd_service.exe crashing.
1165 * Add support for DNS SRV records in place of AFSDB records.Â
1166 _afs3-vlserver._udp.<cellname>. Priority field is used.Â
1167 Weight is currently ignored.
1169 * Add a method of specifying Client CellServDB information within the
1170 registry that can be used to either override the CellServDB file or
1171 force the use of DNS lookups for a given cell.
1173 The registry schema is as follows:
1175 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\
1176 "LinkedCell" REG_SZ "[cellname]"
1177 "Description" REG_SZ "[comment]"
1178 "ForceDNS" DWORD {0,1}
1180 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\[servername]\
1181 "HostName" REG_SZ "[hostname]" Default: [servername]
1182 "IPv4Address" REG_SZ "[address]" Used only if gethostbyname() fails.
1183 "IPv6Address" REG_SZ "[address]" <future>
1184 "Comment" REG_SZ "[comment]"
1185 "Rank" DWORD "0..65535" Default: 0
1186 "Clone" DWORD "{0,1}" <future: server only>
1187 "vlserver" DWORD "7003" <future>
1188 "ptserver" DWORD ... <future>
1190 ForceDNS is implied non-zero if there are no [servername]
1191 keys under the [cellname] key. Otherwise, ForceDNS is zero.
1192 If [servername] keys are specified and none of them evaluate
1193 to a valid server configuration, the return code is success.
1194 This prevents failover to the CellServDB file or DNS.
1196 Only one of "HostName" or "IPv4Address" is required. "HostName"
1197 is optional if [servername] is the host name.
1199 * Extend registry based CellServDB functionality to
1200 afsconf_GetCellInfo() interface used by aklog and many
1201 other command line utilities.
1203 * libafsconf.dll moved from OpenAFS\Client\Program to OpenAFS\Common
1204 as it is now used by both client and server components.
1207 * PriorityClass of afsd_service.exe process raised to
1208 "High" to match the priority of the system services
1209 that are waiting for afsd_service.exe to respond.
1210 Configurable using [TransarcAFSDaemon\Parameters]
1211 DWORD "PriorityClass" value. (use values from Win32
1214 * Change error reporting of CM_ERROR_BADSHARENAME to
1215 use the correct error (NTSTATUS_BAD_NETWORK_PATH).
1216 The error we were sending would cause the smb
1217 redirector to drop the connection.
1219 * Change the error reporting of CM_ERROR_ALLOFFLINE
1220 and CM_ERROR_ALLDOWN to return NTSTATUS_PATH_NOT_FOUND.
1221 Previous we would return Bad Network Path which
1222 would cause the smb redirector to cancel all file
1223 handles on the same "share".
1225 * A debug version of afsd_service.exe can now be used
1226 with a release version of afsrpc.dll (or vice versa).
1227 afsrpc.dll now implements and exports xdr_free().
1229 * Removed all unprotected OutputDebugString() calls as
1230 those calls can block for up to 10 seconds if there
1231 is contention for the globally shared memory or if
1232 the debugger fails to read data in a timely manner.
1234 * Visual Studio 2008, WDK 6.1 and DDK 6.1 are now
1235 supported for builds. The production releases still
1239 * Discovered that XP/2003 supports a variant on recvmsg
1240 called WSARecvMsg. The input parameters are mostly
1241 compatible with the Posix recvmsg. This will save
1242 multiple memcpy() calls per received packet.
1244 * Discovered that Vista/2008 supports a variant on
1245 sendmsg called WSASendMsg. The input parameters
1246 are mostly compatible with the Posix sendmsg. This
1247 will save multiple memcpy() calls per sent packet.
1249 * Exporting rx_ReadProc32, rx_WriteProc32 from afsrpc.dll
1250 and utilizing rx_ReadProc32 from within the cache manager
1251 as part of the FetchData operations. rx_ReadProc32 and
1252 rx_WriteProc32 are optimized versions of rx_ReadProc and
1253 rx_WriteProc for use when reading 32-bit integer values.
1255 * The Network Identity Manager AFS provider now reports
1256 progress events. These provide a better user experience
1257 with Network Identity Manager 2.0.
1259 * [RT 124293] A race condition exists which permits the
1260 scp field of the an smb_fid_t object to become invalid
1261 after a request on the smb_fid_t has begun. Remove the
1262 race by protecting all access to the scp field with the
1263 smb_fid_t mx mutex and obtaining a local reference on
1264 the cm_scache_t object for the length of the request.
1266 * [RT 124276] If the vldb is out of sync with the contents
1267 of the file servers, afsd_service will retry too many
1268 times when a file server reports a volume as not being
1269 present. Now if the list reported by the vldb is the
1270 same as the previously seen list, then the retry is
1273 * [RT 124276] Read-only volume failover was broken in
1274 1.5.53 whenever accessing a volume results in VNOVOL or
1277 * [RT 124276] Prior to 1.3.70 the volume server reference
1278 list was not reference counted and would be prematurely
1279 freed while in use. When reference counting was added
1280 in 1.3.70 a bug was introduced that could result in
1281 service reference list corruption.
1283 * Add Windows Application Event Log warning messages for
1284 "Client SMB MPX value too large" and
1285 "Client SMB Buffer Size too small".
1287 * Renaming of files across directory boundaries would result
1288 in an invalid handle error when attempting to access the
1289 files after the move.
1291 * Fix the handling of Tran2 Set Path Info RPCs. Do not fail
1292 when a smb file descriptor cannot be found. The whole point
1293 of using a Path Info function is because an smb file
1294 descriptor wasn't allocated.
1296 * More edge cases in which dynamic addition of Freelance
1297 root.afs entries would get the wrong FID or where the
1298 root.afs directory would not be refreshed.
1300 * Buffer overflow could occur if the workstation cell name was
1301 longer than 64 characters. Crashes could occur in
1302 afscreds.exe, afslogon.dll, and afsd_service.exe.
1304 * VNOSERVICE and VOFFLINE errors were leaking and were exposed
1307 * Improve consistency of the bad data version value used for
1310 * Log file server uuid values as part of the cm_server object
1311 when available. Dump the cm_server object list in response
1314 * Optimize the performance of resetting access control lists
1315 when tokens are set or removed.
1317 * Fix vos.exe so that it can work with very large volume
1320 * Remove symlink recursion tests and increase max symlink count
1324 * Undo the "UAC manifest fix" applied to afs_config.exe.
1326 * Fixed some additional race conditions in the Rx library
1328 * Ensure that Freelance allocation of vnodes follow
1329 the AFS convention of odd vnodes are directories
1330 and everything else is an even vnode.
1332 * Add Freelance logic to mount point and symlink
1333 evaluation functions.
1335 * Enhance smb_ParseASCIIBlock() so that it can handle
1336 all of the STRING formats defined by the CIFS Technical
1339 * Validate the output of smb_ParseASCIIBlock() in all
1340 callers. Return CM_ERROR_BADSMB if the STRING field
1341 cannot be parsed. CM_ERROR_BADSMB will cause the
1342 contents of the packet to be logged.
1344 * If multiple SMB Raw Write operations were taking
1345 place at the same time, there could be data corruption
1346 because unique event objects were not generated for
1347 each Netbios receive operation.
1351 * Fix a race in rxi_WriteProc which could result
1352 in a freed or in use packet being passed to
1353 rxi_PrepareSendPacket if the call is currently
1354 in a transmit wait state. Passing a freed packet
1355 will result in a panic. Passing an in use packet
1356 might result in a call failure or perhaps data
1359 * Kerberos referrals broke the NIM AFS provider
1360 logic for automated configuration of the workstation
1361 cell with a new identity. Now if a referrals
1362 response is returned, assume the cell belongs to the
1365 * The interface to buf_CleanAsync() and
1366 buf_CleanAsyncLocked() has been changed to permit
1367 errors to be propagated to the caller.
1369 * During an enumeration of a directory, if the btree
1370 has not yet been generated, return an error.
1372 * When mapping CM_ERROR values to NTSTATUS values
1373 do not treat 0 as an error.
1375 * Avoid a cm_scache_t object reference count underflow
1376 when renaming via the SMB interface.
1378 * smb_IsMask() can be called with a non-8.3 input buffer.
1379 Separate smb_IsMask() into smb_Is8Dot3Mask() for 8.3
1380 input buffers and smb_IsMask() for non-8.3 input buffers.
1382 * Add cm_buf_t error state to the "fs memdump" output.
1384 * pioctl drive mapping checks need to take into account
1385 Global Drive Mappings and SUBST assignments.
1387 * Rename operations should not be permitted if the
1388 destination name collides with an existing name that
1389 differs only by case UNLESS the collision is the object
1390 that is being renamed.
1392 * Increase the maximum number of background daemon threads
1393 to 64. Leave the default at 4.
1395 * Set the FILE_UNICODE_ON_DISK flag when reporting the
1396 file system characteristics.
1398 * Report the maximum component name length as 255 even
1399 for Unicode as part of the file system characteristics.
1401 * Add RX debugging information to "fs memdump" output
1404 * Refcount the cm_server_t objects returned by
1407 * Fix the UAC manifest applied to afs_config.exe
1409 * rxdebug now indicates if any RX_CONN flag is set
1410 not just some of them
1412 * Change the mapping of CM_ERROR_WOULDBLOCK from
1413 NTSTATUS_LOCK_NOT_GRANTED to NTSTATUS_CANT_WAIT
1417 * Includes #GCO Public CellServDB 07 Nov 2008
1419 * Prevent separate cell entries from being created for
1420 "foobar.org" and "foobar.org." which can happen if
1421 an AFSDB record is found instead of a CellServDB
1424 * Fix a panic when checking the SMB virtual circuit list.
1425 If a VC is destroyed while the list is being checked
1426 it is possible for the next object to become invalid.
1428 * Increased logging to afsd_init.log of netbios name
1431 * Fix a DNLC panic when a UTF-16 string that cannot be
1432 converted to a utf-8 string is provided as a filename.
1434 * In the NSIS installer, delete old afslogon.dll/afscpcc.exe
1435 from %windir%\system32 and fix the WinLogon\Notify\AfsLogon
1436 location to refer directly to the new location in
1437 %programfiles%\openafs\client\program
1439 * Be more aggressive about responding to Netbios
1440 NRC_BRIDGE/NRC_NOWILD errors. Rebind the Netbios name to
1441 the LANA within the listener thread without performing a
1444 * Use the WNet Enumeration of all connected drives as a filter
1445 in the pioctl GetIoctlHandle operation. This should remove
1446 the last situations in which there was a delay when opening
1447 an Explorer Shell properties menu with the AFS Shell
1450 * Fix the parsing of CellServDB in Windows to handle linked
1451 cells as is done on Unix.
1453 * Add support for linked cells to the cache manager,
1454 netidmgr afs provider, and aklog. When two cells are
1455 linked, a volume lookup that fails in one cell is looked
1456 up in the other. This can be used for construction of
1457 test cells that override specific volumes; cell renaming;
1460 * Increase the length of cell and realm names that can
1461 be entered in the NIM AFS provider.
1463 * Actively prevent a 32-bit afsd_service.exe from
1464 executing in the WOW64 environment.
1466 * Protect against invalid IP address values in CellServDB
1471 * Fix an Rx panic caused by an attempt to double free
1472 an rx_packet when an error occurs while writing the
1475 * Fix an exception cause by a race condition when
1476 processing multiple file change notifications registered
1477 by the Explorer Shell or other applications.
1479 * Modify the Rx debug interface to use var_args and
1480 permit Rx debugging to be used on 64-bit Windows
1481 which previously had it disabled.
1485 * Automated lock hierarchy enforcement added to the
1486 osi base lock package. This permits detection of lock
1487 violations that might but often does not result in
1488 deadlocks. Enabled by default in the checked builds.
1489 May be activated in release builds via the registry
1490 value "LockOrderValidation" in the TransarcAFSDaemon
1493 When a violation is detection an assertion is thrown
1494 generating a dump file and a message indicating which
1495 locks were obtained out of order is logged to the
1496 Windows Application Event Log.
1498 * Attempt to ensure that a dump file will be generated
1499 whenever the service panics.
1501 * Removed unused lock cm_bufGetMutex
1503 * Many out of order lock acquisitions were detected
1504 using the lock order verifier and were corrected.
1506 * A race condition within the function path cm_GetSCache() ->
1507 cm_GetNewSCache() -> cm_RecycleSCache() permitted a
1508 cm_scache_t object to be issued simultaneously to two
1509 threads. This would eventually result in a panic due
1510 to the resulting undercount.
1512 * Fix interpretation of the empty string as the ioctl path
1513 to mean the current directory. "fs lsm", "symlink list",
1514 etc. now return a "not a ..." error instead of "does not
1517 * Add SMB_STRF_SRCNULTERM flag to smb_ParseStringXXX functions.
1518 to indicate that the input string is a nul terminated string.
1519 Assign ti when input strings are nul terminated instead of
1520 length counted. (from 1.5.50)
1522 * The CIFS protocol specification for handling NT_TRANSACT_CREATE
1523 does not match the observed behavor. The 'nameLength' is specified
1524 in bytes, not in characters. Fix the implementation to match
1525 the observed behavior. (1.5.50)
1527 * The cm_HaveAccessRights() attempt at deadlock avoidance by
1528 calling lock_TryRead() on the parent directory cm_scache_t rw-lock
1529 does not avoid the deadlock. Avoid the deadlock by enforcing the
1530 lock order of the lowest vnode first. Then remove the infinite
1531 loop avoidance in cm_SyncOp() that was returning an unwarranted
1532 access denied error.
1534 * File Server lock synchronization was not propertly enforced in
1535 smb_ReceiveNTCreateX() and smb_ReceiveNTTranCreate().
1537 * Another change to the short name generator. Plus and equal are
1538 not valid characters in a short name according to the
1539 RtlIsNameLegalDOS8Dot3() in kernel mode. They have been replaced
1540 with left and right brace in order to minimize the impact.
1542 * Fix a race condition that can result in a heap overwrite error if
1543 the number of known file servers or vldb servers increases during
1544 the preparation phase of a cm_CheckServers() operation.
1546 * If the SMB Authentication would fail, a LSA allocated memory
1547 buffer would be leaked.
1549 * Rewrite the cm_cell_t allocation code to permit cell vldb server
1550 lookups to fail without leaking cm_cell_t objects.
1552 * Add "fs setcrypt auth" to permit the use of rxkad authentication
1553 only mode. Not documented.
1555 * Initialize Rx mutex and lock objects once and only once as part of
1556 the DLL load process.
1558 * When reporting the Volume Name "AFS" do not include the terminating
1559 NUL in the reported length.
1561 * Refresh Volume Location information upon receipt of a VNOVOL error
1562 from the file server and avoid a race condition that prevented the
1563 updated volume location information from being destroyed immediately
1564 after it was acquired.
1566 * In response to VNOVOL and VMOVED, remove the volume id from the server's
1567 volume list so that the volume's up and down status is not affected
1568 when the machine is shutdown.
1570 * Many fixes to the Rx stack related to packet allocation and deallocation;
1571 Rx debugging; Rx packet tracking across queues; and Rx packet loss caused
1572 by overwriting the rx_call currentPacket pointer when it was already
1573 referring to a packet.
1575 * Rx Jumbograms are now off by default. To re-enable, add "RxJumbo" = "1"
1576 in the TransarcAFSDeamon/Parameters key.
1578 * "fs flush", "fs flushvolume", and "fs flushall" now destroy locally
1579 constructed directory B+ trees for flushed directory objects.
1581 * When cached pages for directory objects have been locally modified, do
1582 not permit the locally modified pages to be mixed with pages subsequently
1583 obtained from the file server. Doing so can result in directory
1586 * If Ubix errors are received in response to VL_xxx operations record
1587 the error on the server instance and permit the operation request to
1588 be retried in case another server is able to answer the query. We have
1589 seen at least one instance of a cell in which a VL server returned
1590 UNOQUORUM (possible database truncation) when the other servers were
1591 responding successfully.
1594 * Optimize the assignment of write errors to dirty
1595 buffers by applying the first error to all buffers
1596 belonging to the same object instead of trying each
1599 * Add resource version info to xstat_fs_test and xstat_cm_test
1601 * Fix a delete vs delete[] usage problem in the WiX custom
1604 * When creating a mount point, validate the volume name to
1605 make sure it exists.
1607 * When processing a mount point, pay attention to VL_BADNAME
1610 * Remove a race condition from cm_volume_t recycling that could
1611 have resulted in two threads making use of the same object at
1614 * Fix a reference count leak on smb_vc_t objects introduced in
1615 1.5.50. Permit the objects to be freed once again.
1617 * Fix the output of "fs memdump" for smb_vc_t objects that was
1620 * Add routines for debugging smb refcount errors.
1622 * When shutting down the cache manager, do not permit AFS
1623 callback requests to be processed. Doing so can result in
1624 invalid memory accesses.
1626 * Do not recycle a deleted cm_scache_t object unless the
1627 refCount is 0. Doing so can result in a deadlock and
1628 invalid memory access.
1630 * Minimize the number of RPCs sent to the file server for
1631 directory bulk status requests. Only request status for
1632 objects that we are actually reporting on and only if we
1633 do not already have status for the object and then make
1634 sure that we return as much as we can at once via the SMB
1635 interface but not so much that it exceeds the RDR timeout
1637 * Enable bos restricted operations
1639 * Fix the creation of submounts used by the afscreds and afs_config
1642 * Avoid repeated retries when attempting to obtain access rights.
1643 If the access rights were obtained successfully but do not
1644 satisfy the request, fail the request.
1646 * Fix error code reported when attempting delete a file on a
1647 readonly volume or one that is marked with the readonly DOS
1650 * Fix a short name truncation bug.
1652 * Fix a heap corruption error when reading the CellServDB file
1655 * Fix directory listings when the directory has "lookup" permission
1656 only and the directory has never been previously accessed.
1658 * Fix several tracking issues for cm_scache_t rw locks.
1660 * Ensure that all buffers used to store cell names are the
1663 * Restore Windows 2000 support (broken in 1.5.50).
1664 LOCALE_INVARIANT was introduced in XP.
1666 * Add RxUdpBufSize registry value. Default is 256K.
1668 * Avoid multiple deadlocks between cm_freelanceLock and cm_scacheLock
1670 * Do not include trailing NULs when returning file names in
1671 a directory search response.
1673 * Prevent RX from crashing the afsd_service by attempting
1674 to allocate to many buffers at once when memory is fragmented.
1676 * Avoid a deadlock when freeing cm_server_t objects
1678 * Optimize search cell file. Do not search to end of file
1679 if the cell has already been found.
1681 * Remove explorer shell extension delay on local disks.
1683 * Automatically pre-allocate 64 Rx packets per thread
1685 * Add smb lock requests and stat cache lock allocations to
1686 the output from fs memdump.
1689 * Fix Large File Support (broken in 1.5.50)
1691 * Fix "fs setacl" (broken in 1.5.50)
1693 * Fix pioctl path evaluation of drive letter
1694 mapped paths. (broken in 1.5.50)
1696 * Fix installation of VC8 Runtime library in the EXE
1697 Retail installer. (broken in 1.5.50)
1699 * Avoid a potential deadlock during Volume Location
1700 updates if all the file servers are unreachable at
1701 the time the VL update succeeds.
1703 * Ensure that rx connection object is references
1706 * Fix propagation of over quota errors during cm_FSync()
1709 Since 1.5.39 [1.5.50 released 16 July 2008]
1710 * There were no releases between 1.5.39 and 1.5.50.
1712 * FIRST RELEASE WITH UNICODE SUPPORT
1714 * Pioctl interfaces to the cache manager refactored
1715 to provide layering between the SMB specific code
1716 and the general purpose ioctl operation.
1718 * Garbage collect dead SMB virtual circuits as soon
1719 as they are no longer being referenced. This avoids
1720 problems with outstanding locks not being dropped
1721 when the virtual circuit becomes invalid.
1723 * Remove the IBM Administration Reference documentation
1724 and replace it with the OpenAFS Command Reference
1727 * Avoid calling rx_SetDeadTime and rx_SetHardDeadTime
1728 functions each time a connection is about to be used.
1729 Do not hold a lock on the rx connection object while
1730 it is being selected. This avoids a race between
1731 threads attempting to set the timeout values and
1732 removes a bottleneck that was hampering performance.
1734 * Ensure that the smb directory attribute is set
1735 for all directory objects.
1737 * Add vs2008 support to the NSIS installer scripts
1739 * Replace the VC Runtime EXE installer with the MSI
1740 installer in the NSIS installer scripts
1742 * Properly delete the VC7.1 C Runtime libraries.
1744 Since 1.5.36 [1.5.39 released 23 June 2008]
1745 * There were no 1.5.37 or 1.5.38 releases for Windows
1747 * Use Visual Studio version of vsprintf() instead of
1748 home grown version that resulted in linking conflicts
1750 * Prevent buserver, ptserver and fileserver from crashing
1751 due to non-portable strftime() format strings.
1753 * Prevent server crashes caused by rx_InitMorePackets()
1754 being called prior to allocation of all of the required
1755 synchronization objects.
1757 * Fixed a memory leak in the internal B+ tree directory.
1758 For objects with non-8.3 names, the 'longname' memory
1759 allocation would be leaked during object deletion or
1762 * Fixed a memory leak in the background daemon cm_CheckServer()
1763 routine. The memory allocated to store the server list
1766 * Obtain a missing lock around a call to cm_RemoveSCacheFromHashTable().
1768 * Correct an abstraction layer violation. cm_scache_t objects
1769 should be marked deleted in cm_Unlink() and cm_RemoveDir() and not
1770 in smb_CloseFID(). Cleanup of deleted cm_scache_t objects should be
1771 performed in cm_ReleaseSCache() when the reference count hits zero.
1773 * Prototype cm_AdjustScacheLRU() and re-implement it using osi_QAddH().
1775 * Do not reference 'smb_logp' in cm_vnodeops.c. Use 'afsd_logp'
1778 * Ensure that scache object 'nextp' pointers are set to NULL
1779 when the objects are removed from the hash table linked list.
1781 * If a dirty buffer cannot be written to the file server because
1782 of an access denied error, mark the error in the cm_buf_t
1783 structure so that it can be dealt with instead of retrying
1787 * Update CellServDB file to GCO Public 23 Apr 2008
1789 * Fix a cm_buf_t reference count leak when attempts to write
1790 dirty buffers to the file server from within cm_IncrSyncer()
1793 * Prevent udebug from crashing.
1795 * Activate RX Idle Timeouts. If the file server is busy for more
1796 then 30 seconds, attempt to failover to another server without
1797 marking the busy server down.
1799 * Another VNOVNODE issue fixed. When writing a dirty buffer
1800 to the file server, if VNOVNODE is received, mark all buffers
1801 as invalid without further attempts to contact the file server.
1803 * Improved performance on high latency links.
1806 * Fix a bug preventing the re-initialization of the Freelance
1807 root.cell contents when a change is made.
1809 * Properly return CM_ERROR_NOSUCHFILE when removing a symlink
1810 or mountpoint that does not exist.
1812 * aklog, afscreds, integrated logon, and the NIM AFS credential
1813 provider will all attempt to first acquire an afs service ticket
1814 using the service principal: afs/<cell>@<CLIENT_REALM>. This
1815 permits the proper detection of the cell realm when multiple
1816 realms are local realms for the cell.
1818 * Fix fs commands that re-use file identifiers instead of forcing
1819 a path evaluation each time. This was broken in 1.5.33 when
1820 hashes were added to the cm_fid_t structure.
1822 * Fix symlink and mountpoint removal and then recreation when
1823 applied to the fake Freelance root.afs volume.
1825 * Fix SMB change notifications for the Freelance root.afs volume
1826 when creating or removing symlinks and mount points.
1828 * Force DNS AFSDB cell name lookups to lower case if the DNS
1829 response is not all lower case.
1831 * 32-bit installers are now built with Visual Studio 2005
1833 * All vos.exe commands now support the "-noresolve" option which
1834 forces the display of IP addresses instead of DNS hostnames.
1836 * No longer hold the cm_volume_t mutex across VL_xxx RPCs. This
1837 permits operations that do not require up to date volume
1838 location data to continue while waiting for the RPC to complete.
1839 Multiple threads waiting for a volume location update on the
1840 same volume set no longer trigger multiple calls VL_xxx RPCs.
1842 * Volume server references could be marked offline and never
1843 reset to the not_busy state.
1845 * Further optimizations to cm_Analyze() retry logic for
1846 CM_ERROR_OFFLINE and CM_ERROR_ALLBUSY error handling.
1849 * Optimize the DNLC by applying Interlocked incrementing for
1850 statistics gathering and enabling greater use of read
1851 locks instead of write locks
1853 * Further optimize osi_Log operations when logging is disabled
1854 by removing the need for the function call overhead.
1856 * Add a new lock operation, Convert Read to Write, and apply
1857 it throughout the cache manager. This function provides a
1858 fast transition from a read lock to a write lock when the
1859 caller is the only reader.
1861 * Fix a timeout problem when opening files or request locks
1862 when the file in question is already write-locked by another
1865 * Fix a deadlock that can occur if "fs flushall" is executed
1866 while applications are writing to AFS.
1868 * The cache manager would re-read data from the file server
1869 after file truncation when obtaining buffers to write to the
1870 same file until the first StoreData RPC completes.
1872 * Fixes error handling in cm_NTCheckDelete()
1874 * Replaces cm_scache_t mutex with a read-write lock permitting
1875 additional parallel access.
1877 * Prevent an cm_scache_t reference undercount when evaluating
1878 symlinks containing @sys when none of the active sysnames
1879 match any of the available targets.
1881 * Prevent the leak of a cm_scache_t rwlock if associated symlink
1884 * Set RxMaxMTU registry value to 0 which is the equivalent of
1885 it not being set at all.
1887 * Reduce contention between BkgDaemon threads
1889 * Prevent CM_SCACHEFLAG_ASYNCSTORE from being reset on a write
1892 * Convert reference counts on cm_volume_t, cm_conn_t, cm_server_t
1893 objects to use the faster Interlocked operations which avoid
1894 the need to hold write locks.
1896 * Permit rxkad errors other than RXKADEXPIRED to be treated as a
1897 non-fatal error that forces a retry against another service
1900 * Fix a crash in smb_WriteData that would occur if an application
1901 opened the file as read-only and then sent a write request anyway.
1904 * Respond to VL_NOENT errors by removing cm_volume_t object and
1905 setting it to be the first object to be recycled. This is not
1906 only a more efficient use of resources but it also prevents
1907 repeated attempts to query the VLDB server for the volume from
1908 the CheckOfflineVolumes routine.
1910 * Change the defaults for RxEnablePeerStats and RxEnableProcessStats
1913 * A first cut at a cache manager statistics monitor that can be used
1914 to determine the necessary cache parameters to support the
1915 application working set.
1917 Off by default, the performance package can be activated by setting
1918 "daemonPerformanceTuningInterval" in the service Parameters key.
1919 As with the other daemon interval values the unit is in seconds.
1921 At service start and each succeeding interval the cache manager
1922 will write statistics to %TEMP%\afsd_performance.txt showing the
1923 relative usage of cm_scache_t, cm_volume_t and cm_buf_t objects.
1924 The FID statistics keep track of all FIDs seen by the cache manager
1925 during the service session whether or not they are backed by any
1926 live object in the cache.
1928 These statistics are not stored in the cache file.
1930 * Fix windows event logging broken in 1.5.33.
1932 * Eliminate multiple calls to syscfg_GetIFInfo() which is used to
1933 populate the IP address list for WhoAreYou/TellMeAboutYourself RPCs
1934 as well as when computing the server preference list.
1936 * When merging status after a StoreData operation prevent data
1937 buffers from being discarded that we would prefer to keep.
1938 (Bug introduced in 1.5.33.)
1940 * Modify installers the set RxMaxMTU to 0.
1942 * In the AFS Configuration Panel on the Advanced->Miscellaneous
1943 dialog, the Background Daemons and Service Threads fields were
1946 * Increase default number of Background Daemons to 4.
1948 * Remove cm_volume_t reference from the cm_scache_t object. Permit
1949 cm_volume_t objects to be recycled.
1951 * Windows Error Reporting has shown a number of previously unseen
1952 stack corruption errors in the Explorer Shell extension. The
1953 minidump appeared to indicate the problem was in ParseAcl().
1954 Reviewing the function it appears that numerous things could go
1955 wrong if invalid input was provided. Added error checking and
1956 data validation. Hopefully this will address the problem.
1958 * Windows Error Reporting has shown crashes in aklog as a result
1959 of krb5_cc_get_default() failures not being handled. They are
1963 * The Rx library used a 32-bit type for sockets which was
1964 truncating the socket value on 64-bit Windows. This
1965 was not an issue in general but became an issue when
1966 Rx "hot thread" support was enabled.
1968 * Enable "Rx hot thread" support which ensures that while
1969 a received message is being processed that a new thread
1970 is allocated to receive the next message to arrive.
1972 * Increased parallel access to smb_rctLock protected
1973 resources. Reduce the number of times that the
1974 lock is dropped and re-obtained.
1976 * Increased parallel access to cm_scache_t and cm_buf_t
1977 resources by replacing the previous global write lock
1978 protected reference counting mechanism with one built
1979 upon the use of InterlockedIncrement and InterlockedDecrement.
1981 * Implement a hash for cm_fid_t to reduce comparison costs
1982 and inline the cm_FidCmp() function.
1984 * Redefine the BUF_HASH and BUF_FILEHASH in terms of the cm_fid_t
1985 hash which has a better distribution
1987 * Modify cm_ConsiderPrefetch to evaluate the amount of data in
1988 the most recent read request instead of the next chunkSize.
1989 cm_chunkSize can be dozens or hundreds of buffers. As a result
1990 too much time is spent performing the evaluation.
1992 * Fix the usage of cm_scache_t bufCreateLock. The purpose of this
1993 lock is to prevent the creation of new buffers while a truncation
1994 is being performed. All references to bufCreateLock have been
1995 removed except in two places:
1996 i. a write-lock surrounding the function that calls buf_Truncate()
1997 ii. a read-lock within buf_GetNewLocked() that actually allocates
2000 * Modify the CM_CONFIG_DATA_MAGIC value to include a
2001 CM_CONFIG_DATA_VERSION number which value be used to force the
2002 replacement of the cache file contents when incompatible changes
2003 are made between releases.
2005 * CM_SCACHESYNC_ASYNCSTORE should not be ordered by
2006 cm_SyncOpCheckContinue
2008 * Avoid calls to multi_Rx if nconns == 0
2010 * Modify smb_WriteData to perform background writes based upon the
2011 crossing of 'smb_AsyncStoreSize' boundaries instead of cm_chunkSize
2012 boundaries. This will slow down writes from the SMB interface but
2013 will avoid the risk of the CIFS client disconnecting from the AFS
2014 client SMB server. The default value for 'smb_AsyncStoreSize'
2017 * SMB AsyncStore functionality is on by default but can be
2018 disabled using the registry value:
2020 HKLM\SOFTWARE\OpenAFS\Client DWORD "EnableSMBAsyncStore"
2022 smb_AsyncStoreSize can be adjusted with
2024 HKLM\SOFTWARE\OpenAFS\Client DWORD "SMBAsyncStoreSize"
2026 The value must be a multiple of the buffer block size
2027 and cannot be larger than the chunk size.
2029 * Do not leak a read lock on the bufCreateLock
2031 * Do not forget to call cm_SyncOpDone if cm_SyncOp succeeded
2034 * In MergeStatus, only remove clean buffers from the hash table.
2036 * Avoid the need to update the data version number on each buffer
2037 associated with a scache when MergeStatus is called after a
2038 StoreData by maintaining a range of valid data versions as part
2039 of the cm_scache_t object.
2041 * Profiling shows large numbers of blocked calls in cm_HoldSCache
2042 from cm_HaveAccess... We can safely avoid the call so do so.
2044 * Add "ForceAnonVLDB" service parameters registry value. Default
2045 is 0. Set to 1 to force the use of anonymous connections to
2046 the VLDB servers. This is useful when using krb5 tokens and the
2047 VLDB servers have not been updated.
2050 * Do not attempt to search for a cell name that matches one
2051 of the special share names "svrsvc", "wkssvc", "ipc$"
2053 * Issue permission denied warnings from afscreds if attempts
2054 to modify the startup shortcut or registry values fail.
2055 This happens under Vista UAC.
2057 * Do not return BAD_NETWORK_ERROR in response to attempts
2058 to create directories or files in the Freelance root.afs
2059 volume. Instead return ACCESS_DENIED.
2061 BAD_NETWORK_ERROR was being sent because the Freelance cell
2062 has no vldb servers. Short circuit the error by testing for
2063 freelance in CreateFile and MakeDir
2065 * Treat srvsvc, wkssvc and ipc$ as pseudo files. Permit them
2066 to be opened and closed without actually existing. For now
2067 we treat them as equivalent to ioctls but this should
2070 * Fix RAP processing to not reject names that exist in the
2071 root.afs volume for GetInfo queries.
2073 * In afscreds, use GetUserNameEx to obtain the username instead
2074 of searching the registry. Try UserPrincipalName first and
2075 fallback to SamCompatible.
2077 * Stop linking to rpcns4.dll which is no longer supported on
2078 Vista and 2008. TaAfsAdmSvr is going to have to be
2081 * When processing dfs referral requests, if the sharename is
2082 a partial match and therefore will not be added automatically
2083 to the share list, make sure that we return an error.
2085 * Avoid a crash when logging VNOVOL errors returned from
2086 cm_CheckOfflineVolume.
2088 * When trace logging is disabled, do not permit osi_LogSaveString
2089 to perform the string copying. By disabling the string
2090 copies, throughput of the service more than doubles.
2092 * when checking offline volumes, update the cell vlserver info
2093 if required and don't perform a check if there are no servers
2096 * Add volumes in the vl_unknown state to the list of objects to be
2097 checked as offline or retried when an allbusy error is received
2100 * Add support for configuring rx_ExtraPackets
2102 * All calls to buf_Get() must be protected by a read lock on the
2103 cm_scache_t bufCreateLock
2105 * When a MergeStatus results in the invalidation of buffers, do
2106 not remove buffers from the hash tables that have active references.
2108 * Upon receiving a VNOVNODE error from a file server, be sure to
2109 invalidate any file locks. This prevents on-going attempts to
2110 extend the locks and will cause subsequent attempts to access
2111 the file to receive a invalid handle error.
2113 * Prevent smb_ReceiveTran2SearchDir from adding terminating NULs
2114 to directory entry names.
2116 * No longer discard tokens when RXKADBADTICKET error is received.
2117 Instead return STATUS_NO_KERB_KEY error.
2120 * Add "cmdebug <host> -cellservdb" option and client service
2121 implementation permits querying the vldb server for each
2122 cell known to the cache manager. The output is in a form
2123 that permits it to be used as a CellServDB file.
2125 * Add Symlinks->Show option to the AFS Explorer Shell Context
2128 * Prevent buffer overrun in the cmd parser used by command
2129 line tools such as fs, vos, pts, etc.
2131 * The path ioctl operations have several issues:
2133 (1) the specified path for "fs examine, whereis, whichcell,
2134 flush" and so always has the follow symlinks and mount
2135 points semantics. This makes it impossible to determine
2136 what the FID of a symlink or mount point is.
2138 (2) "fs examine" out is not the result of a single pioctl
2139 operation but is actually the combined output of half
2140 a dozen operations. Path evaluation is an expensive
2141 operation. It would be faster if the caller could
2142 evaluate the FID first and then perform all of the
2143 rest of the operations by specifying the FID instead
2146 (3) fs output reports all objects as files. By adding a
2147 GetFileType pioctl more informative output can be provided
2148 that indicates what type of object the path evaluates to.
2150 (4) the Windows fs command includes a number of commands that
2151 do nothing but exist only because the Unix cache manager
2154 A new extendible data structure cm_ioctl_query_opts_t has been
2155 added which can be optionally specified with pioctls that do not
2156 already require input data. The first two fields of this structure
2157 are 'literal' and 'fid'. The literal field is used to indicate
2158 whether the last component of the path should be evaluated following
2159 symlinks and mount points.
2161 The fid field permits a fid to be specified.
2163 * A new GetFileType pioctl has been added. The type of objects are
2166 * A new "-literal" option is available for "fs examine, flush, whereis,
2169 * Unimplemented fs commands have been removed.
2171 * There are circumstances where a volume object is being accessed and
2172 the volume is marked indicating that the volume location information
2173 is out of date but where it is also pointless and perhaps dangerous
2174 to block waiting for the rpc to complete. One example is when
2175 processing the cmdebug requests. If we know that we are not going
2176 to use the volume object to contact a server, then we can now set
2177 the CM_GETVOL_FLAG_NO_RESET flag.
2179 * when moving up in the directory tree we search the recorded fid list
2180 to find a matching fid that we have already crossed. we must also
2181 reset the fid count based upon what we discover.
2183 * Cell names are published as share names. As a result they are searched
2184 for as part of the Dfs Referral evaluation. If share "foo" can not be
2185 reached, the CIFS client will fallback to searching for "fo" as well.
2187 Since the freelance client automatically adds a symlink for prefixes
2188 of the cell name, this results in both "foo" and "fo" being added to
2189 the freelance root.afs volume. This patch prevents that negative side
2192 * when tracking Kerberos credential cache names be sure to include
2193 the cache type prefix.
2195 * Microsoft has assigned OpenAFS a network type value
2197 * The algorithms that were used to produce a Netbios name were broken.
2198 The name that was produced when the hostname was too long was in fact
2199 longer than the maximum permitted netbios name.
2201 Also, the "NetbiosName" value was not used as a suffix when the loopback
2202 adapter was not installed. The hardcoded string "AFS" was used instead.
2204 * avoid another deadlock during server probes initiated by the ipaddr
2205 change daemon thread.
2207 * prevent the afs client service from crashing when fs newcell is
2210 * In the NSIS installer, move afslogon.dll, afscpcc.exe and afs_cpa.cpl
2211 to \Program Files\OpenAFS\Client\Program in order to get them out
2212 of the %windir% tree
2214 * EDQUOT == WSAEDQUOT. Define it in the right places so that in the
2215 end STATUS_OUT_OF_QUOTA can be returned to the application.
2217 * Modify the search order for determining the location of CellServDB
2218 and other client configuration files.
2220 1. AFSCONF environment variable
2222 3. NEW - use All Users\AppData\OpenAFS\Client only if CellServDB exists
2223 4. use Program Files\OpenAFS\Client
2225 * More Vista Power Management tweaks
2227 * Close a socket handle leak in the kauth routines
2229 * Don't hold a cm_volume_t mutex across RPCs
2231 * Convert downTimes to use clock time instead of relative times.
2232 This permits correct comparisons with expiration times.
2234 * Update the Wix installer to properly install the prerequisites
2235 for translate_et.exe
2237 * Do not translate pioctl paths as they are not converted by the
2240 * Implement multi_rx probes for checking whether servers are up
2241 or down. Turns probes into a constant time activity instead
2242 of one dependent upon the number of known servers.
2244 * Treat VNOVOL errors the same as VOFFLINE. Do not force an
2245 update of the volume location data and retry. VNOVOL means
2246 that the volume could not be attached. It doesn't mean that
2247 the volume is not on this server.
2249 * Modify the Explorer Shell AFS Context menu Symlink->Add to
2250 not attempt to validate the target path. There is no reliable
2253 * Improve performance of \\afs\<cell><mount-type><volume>\
2254 access by eliminating extraneous attempts to search the
2255 CellServDB file and DNS AFSDB records. Share names
2256 containing '%' or '#' are not valid cell names.
2258 * Return STATUS_RANGE_NOT_LOCKED if the application attempts
2259 to unlock a range that is not currently locked.
2261 * Enable the 32-bit tools to read the 64-bit registry keys
2262 when executing in the WOW64 environment.
2266 * Fix buffer refcount leak introduced in 1.5.29
2268 * Prevent the NIM AFS Provider from crashing when configured to use
2269 Kerberos v4 for token acquisition in situations where no Kerberos v4
2270 support is available. For example, 64-bit KFW or sites that delete
2271 the Kerberos v4 DLL.
2273 * In afsd_init.log, support logging cache sizes greater than 2GB.
2275 * The volume status output of "fs examine" was being determined based
2276 upon the wrong error value. Instead of using the pioctl() return
2277 code, the errno value must be used.
2279 * Prevent invalid pts auto-registration attempts by aklog and the
2280 NIM afs provider when Kerberos referrals are in use. (krb5 1.6+)
2282 * Add synchronization protection to all of the SMB Listener State
2283 variables. This prevents race conditions when the addition or
2284 loss of a network adapter takes place.
2286 * On Vista, there is a race condition between the restart of the
2287 afs client service and the network adapters. Ensure that the
2288 afs client service can handle switching between loopback and
2289 non-loopback modes of operation.
2291 * Fix FollowBackupPath to work with mount points that are not
2292 explicitly read/write.
2294 * Fix a deadlock that has been observed on Vista while resuming
2295 after at least four hours of sleep.
2297 * Add new HKLM registry configuration options that can be used
2298 to provide configuration hints to the NIM AFS Provider when
2299 creating new identities. See release notes for details on
2300 HKLM\SOFTWARE\OpenAFS\Client\Realms\.
2303 * Speed up write buffer operations by holding a global read-lock
2304 instead of a write-lock when updating the buffer data version
2305 number. The data version is protect by the associated stat
2308 * Speed up read buffer operations by removing out of date buffers
2309 from the buffer hash tables.
2311 * Protect against a null pointer dereference in afslogon.dll
2312 if QueryAdHomePathFromSid() fails.
2314 * Remove a race condition while accessing private freelance data
2317 * Update the buffer data versions when Freelance directory data
2320 * Complete the implementation of smb_ReceiveTran2GetDFSReferral().
2323 * Return EAS_NOT_SUPPORTED in response to EA read/store/query
2324 instead of returning zero length.
2326 * Return ERROR_RETRY when a server sends VBUSY.
2328 * Return PATH_NOT_COVERED instead of BADSHARENAME for DfsLinks when
2329 a volstat plug-in is in use.
2331 * Enhance the volstat interface to support Dfs path mappings.
2333 * Fix "symlink list" to do the right thing for Dfs links
2335 * Fix callback expiration that was broken in 1.5.27.
2337 * Fix a crash at shutdown caused by the lack of synchronization
2338 within the daemon threads (cm_Daemon, cm_BkgDaemon, cm_IPAddrDaemon)
2340 * Hold the cm_server_t mutex while walking or modifying the
2341 cm_server_vols_t list in order to avoid a race condition.
2343 * Ensure that the cm_server_t refCount is safely modified.
2345 * The pthread library implementations of pthread_cond_wait
2346 and pthread_cond_timedwait were failing to return with
2347 the mutex held when error conditions occurred. This could
2348 result in race conditions wherever they are used. One case
2349 in which problems were seen was the Rx event_handler().
2351 * Prevent cm_GetBuffer() from locking all buffers associated with
2352 a stat cache object before deciding if the current buffer is up
2355 * Fix cm_BkgPrefetch to actually ensure that there is a cm_buf_t
2356 object for each offset that we will attempt to reserve.
2358 * Reduce the default number of background daemon threads used to
2359 prefetch or store to one in order to prevent two or more threads
2360 from stepping on each other.
2362 * Reduce the default chunk size to 256KB in order to prevent read
2363 timeouts on low bandwidth connections.
2365 * Fix a race condition that could result in threads becoming stranded
2366 when synchronizing multiple requests on the same stat cache object.
2369 * The GiveUpAllCallBacks functionality that was added in 1.5.21
2370 has been disabled. It has been disabled because it turns out
2371 that the RPC is improperly implemented in all file servers
2372 previous to 1.4.6. The RPC is executed without the proper
2373 locks being held resulting in data corruption and eventual file
2374 server failure. The functionality will be re-enabled in a
2377 * a couple of .readonly Volume Callback optimizations were
2379 - Apply most recent volume callback to all stat cache objects
2380 in the volume. This avoids premature callback expirations
2381 and unnecessary FetchStatus calls.
2382 - Add option to permit attempts to renew .readonly Volume
2383 Callbacks on a periodic basis to avoid invalidation of stat
2384 cache data. [Requires daemonCheckVolCBInterval be set to
2385 non-zero number of seconds. 1800 (30 minutes) is suggested.]
2387 * Adds support for 64-bit data version values. Previous releases
2388 only supported 32-bit values. Larger values sent by the file server
2389 would result in rollover.
2391 * aklog and asetkey no longer generate an exception if Kerberos for
2392 Windows is not installed.
2394 * If the LSA authentication functions report a temporary out of memory
2395 condition, fail the current authentication and do not cause the AFS
2396 client service to terminate unexpectedly.
2398 * When the AFS client service is shutting down, prevent the daemon check
2399 thread from executing additional checks which might attempt to access
2400 discarded locks or unmapped memory.
2402 * Correct a double free condition in the NIM AFS credentials provider.
2404 * Add "FollowBackupPath" service parameter registry value.
2405 This provides equivalent functionality to the UNIX afsd -backuptree
2406 option. When set to a non-zero value, normal mount points originating
2407 in a .backup volume will prefer a .backup volume.
2410 * Fix NSIS uninstall from the .MSI installer.
2412 * Add Stack Frame logging for AMD64
2414 * Add support for EWOULDBLOCK to cm_Analyze. If the file
2415 server returns EWOULDBLOCK, retry the request every two
2416 seconds for up to the RDRtimeout.
2418 * Reorganize the order of the includes to ensure that EWOULDBLOCK
2419 is not assigned the same value as EIO.
2421 * Update CellServDB to GCO Public 25 Oct 2007
2423 * Add additional validation and error handling code after
2424 each call to getSlot(). If an invalid slot is returned,
2425 return NONODE. If the invalid slot is returned when
2426 extracting a data node, invalidate the tree.
2428 * Modify compareKeys() to always perform a case-insensitive
2429 comparison and only perform a case sensistive comparison
2430 if the case-insensitive one matches. This ensures the
2431 ordering is consistently reported.
2433 * Add lock assertions to ensure that all calls are being
2434 performed with the correct locks being held. There have been
2435 some crash reports that provide stack data that does not appear
2436 to be possible unless there is a race. However, there are no
2437 obvious locations where the race is taking place and the test
2438 suite indicates that all of the correct locks are being held.
2439 We shall see what happens in the field.
2441 * For consistency replace all calls to findKey in which the range i
2442 s (1,numentries) with calls to getSlot().
2444 * Optimize the depth search loop by testing the slot value in the
2445 for statement instead of forcing the loop to be broken later.
2447 * Reorganize the locking for cm_BeginDirOp and cm_EndDirOp.
2448 There are a number of locations where locks are obtained, dropped,
2449 and reobtained. This reorganization attempts to accomplish several
2452 (1) be optimistic for the most common case so it will be fast
2454 (2) add consistency checks after each location where locks are
2455 dropped and re-obtained. If we lose a race in cm_BeginDirOp
2456 and the bplus tree is out of date, retry until we get to a
2457 consistent state that we can use.
2459 (3) Ensure that all operations take place with the correct locks.
2461 * One of the issues that has become a serious problem since the
2462 addition of the local directory updates is that although cm_SyncOp
2463 synchronizes operations, it does not preserve the order of requests.
2464 This has always been a problem in that it has been possible for a
2465 request to fail to complete due to its worker thread's bad luck.
2466 When a request takes longer than the Windows SMB Redirector's timeout,
2467 the SMB Redirector tears down the SMB virtual circuit.
2469 When using the local directory updates it is really important that
2470 the directory update operations complete in the order that they were
2471 sent to the file server. If they don't, then the local directory state
2472 and the file server state will not match and the local directory state
2473 must be discarded which in turn forces a new read of the entire
2474 directory contents over the network.
2476 This patch adds a new cm_scache_waiter_t object that is used to store
2477 the current thread, buffer, and syncop flags within a waiters queue
2478 on each cm_scache_t object. If a thread is forced to sleep in
2479 cm_SyncOp, upon waking it will check to see if there are any other
2480 threads waiting that are attempting to perform a similar task ahead
2481 of it in the queue. If yes, the thread goes back to sleep. If not,
2482 it goes ahead and enters the cm_SyncOp conflict resolution block.
2484 This patch has the additional side effect of reducing the number of
2485 competing threads that must obtain the cm_scache_t mutex and process
2486 the cm_SyncOp conflict resolution block. As a result, the overall CPU
2487 utilization of the service and the clock time associated with processing
2488 requests will be reduced.
2490 * assert that the cm_scache_t mutex is held when calling
2491 buf_ForceDataVersion().
2493 * rename findNode to leafNode in bplus_Lookup
2495 * replace all OutputDebugString calls with osi_LogX calls
2497 * modify bestMatch to special case the return values for leaf nodes.
2498 If an entry is above or below the values available in the leaf node
2499 return BTLOWER or BTUPPER instead of BTERROR.
2501 * In insert and delete operations check for BTLOWER/BTUPPER and isleaf,
2502 if true convert to either slot 0 or Max and perform the insertion.
2503 This produces easier to read code when performing lookups.
2505 * modify lock_AssertXXX macros to call osi_assertx() and provide a
2506 descriptive message.
2508 * the difference between osi_assert() and osi_assertx() is whether or not
2509 a description is specified as part of the call. When no message is
2510 specified afsd_notifier() is called with a NULL msgp parameter. This
2511 results in a null pointer dereference during the EventLog() call.
2513 As a result, none of the information describing the assertion is written
2516 This commit sets a default message for use when no other message is
2517 provided and it replaces all calls to osi_assert with osi_assertx and
2520 * Migrate B+ tree search key into thread local storage
2523 * Fix an uninitialized variable in lana_IsLoopback() which
2524 could result in a random determination that a loopback
2525 adapter is not a loopback adapter. This would result in
2526 a computed Netbios service name of %MACHINE%-AFS instead
2527 of AFS. This is a serious problem if the value computed
2528 by the AFS Client Service and the client process pioctl
2529 calls produce different results as it will result in a
2530 communication failure.
2532 * Fix a crash in the B+ Tree code that results when a file
2533 rename only results in a change of case.
2535 * Fix the pioctl path parsing. The previous algorithm
2536 broke the current directory of the client into two parts,
2537 the TIDpath which represented the device root and the
2538 relative path. The problem is that this is not sufficient
2539 from proper processing of symlinks and mount points.
2541 What is required is that the last component of the path
2542 must not have the symlink or mount point be evaluated
2543 whereas symlink and mount point evaluation must be performed
2544 for the rest of the relative path. Therefore, the path is
2545 now split into three parts: TIDpath, intermediate path,
2548 "fs examine" now reports the data of the symlink or mount
2549 point object instead of the target and "symlink list" and
2550 "symlink remove" now works. The "Explorer Shell Extension"
2551 menu now is displayed for symlink objects.
2553 * More informative symlink.exe error messages.
2555 * When the cmd.exe "del" command is used to remove a symlink,
2556 not only was the link object removed, but the stat cache
2557 object for the target was marked deleted. This resulted
2558 in cache corruption which would require the destruction
2559 of the cache to correct.
2561 * Add undocumented registry value, "OfflineReadOnlyIsValid"
2562 which results in cached contents from read only volumes
2563 to be treated as valid when there is no callback registered
2564 and all servers are down or all instances are offline.
2566 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
2568 * Add documented registry value, "AcceptDottedPrincipalNames"
2569 which is used to permit the acquisition of AFS tokens
2570 using Kerberos v5 principal names which include a dot in
2571 the first component.
2573 {HKLM\HKCU}\Software\OpenAFS\Client
2575 * "fs uuid" does not require "AFS Client Administrator" group
2576 membership to display the current UUID value.
2579 * Update NSIS installer to version 2.30
2581 * Update Wix installer to version 2.0.5325
2583 * 64-bit KFW support:
2584 - Network Identity Manager plug-in
2586 - 64-bit krb5 enable afscreds.exe, afs server manager,
2589 * Enable B+ tree Directory Search by default.
2591 * NIM plug-in is now compatible with FILE: ccaches
2593 * Install instloop.exe into Client/Programs/ even though
2594 it isn't required. Permit it to be used by organizations
2595 that clone Windows systems.
2597 * On a suspend, only drop callbacks for servers that
2598 are online. If the server is offline, maintain the callback
2599 until the server is once again available.
2601 * Prevent against acquisition of a directory write lock when
2602 there are no updates to be performed. This was resulting
2603 in a write-lock leak which later on would lead to a deadlock.
2605 * Add configurable file pre-fetch mode. For files that match
2606 the specified file extensions, pre-fetch the entire file into
2609 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2610 (MULTI_SZ) PrefetchExecutableExtensions
2612 * Fix the client_osi large integer support so that return values
2613 can be properly assigned to large integer variables.
2615 * Do not hold cm_scache_t mutex across cm_GetCell() calls
2616 since cm_GetCell() can block in a RPC.
2618 * Add short file name support to B+ tree searches.
2620 * Resolve a conflict between the desire to notify file servers
2621 that the client workstation is suspending and the desire to
2622 permit objects with valid callbacks when the client suspends
2623 to continue to use those objects from the cache if the client
2624 resumes without network connectivity.
2626 * Fix stack variables to ensure that all buffers storing paths
2627 can handle AFSPATHMAX (1024) characters. Paths of 256 characters
2628 would cause a stack overflow condition.
2630 * On Vista, force afs_config.exe to run with the highest privileges
2631 available to the end user. For a user in the Administrators
2632 group running under UAC, this will cause afs_config.exe to be
2633 be run with full Administrators privileges after the user
2634 approves the privilege escalation.
2636 * More VNOVNODE handling issues. If a dirty buffer cannot be
2637 written due to VNOVNODE, clear the buffer and ignore the error.
2638 If a cm_scache_t object has been marked deleted due to a VNOVNODE
2639 error, the next time the smb_fid_t object that references it is
2640 used return either NOSUCHFILE or NOSUCHPATH and close the
2641 smb_fid_t object. Subsequent attempts to use the smb_fid_t will
2642 result in an invalid handle error.
2644 * For Vista, set the RequestExecutionLevel of the NSIS installer
2647 * For the build environment, support the generation of manifests
2648 when using the CL1310 (VS .NET 2003) compiler/linker.
2650 * There was a bug in OutputDebugHexString() which could overflow
2651 a stack buffer resulting in unexpected process termination
2652 of afsd_service.exe if the input Security Token contained
2653 many bytes equivalent to the '%' char.
2655 * Add directory lookup statistics to "fs memdump" output.
2659 * Windows uses case-insensitive file name pattern matching
2660 but AFS is a case sensitive file system. The AFS3 directory
2661 format is block based, uses network byte order and
2662 includes a hash table for fast case sensitive lookups.
2663 This causes several problems for the Windows AFS client.
2664 (1) Traversing the directory blocks is cpu expensive
2665 (2) A hash table miss does not indicate that the desired
2666 entry does not exist.
2667 (3) Determining whether a non-ambiguous inexact match or
2668 the entry does not exist requires a linear traversal
2669 of the entire directory.
2670 These issues often result in 100% CPU utilization.
2672 These issues are addressed by building a modified B+ tree for
2673 each directory and then using the B+ tree for searches.
2675 This feature is disabled by default. To enable it set
2676 the "BPlusTrees" registry value to 0x01 in the AFS Client
2677 Service Parameters key.
2679 * Correct a null pointer dereference in the symlink recursion
2680 detection code. (Introduced in 1.5.22)
2682 * Add the Network Identity Manager AFS Provider to AMD64
2687 * A new registry value, "BlockSize", can now be used to adjust
2688 the size of the internal buffers used to store file data.
2689 The default is 4KB. When set, the values 1..1024 specify
2690 multiples of 4KB. Values > 1024 specify actual block sizes
2691 and must be multiples of 4KB. The actual block size in KB
2692 can not be larger than the chunk size which defaults to 1MB.
2694 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2697 * pts.exe and ptserver.exe built with -DSUPERGROUPS
2699 * Service Manager shutdown hints have been added. The hint
2700 specifies that it may take the AFS client service up to two
2701 minutes to shutdown.
2703 * Fix the recursion detection feature introduced in 1.5.22.
2704 The previous implementation broke the evaluation of paths
2705 using ../path or ./path. It also broke the evaluation of
2706 symlinks to absolute paths. For example, /afs/cell/path.
2708 * Fix an uninitialized variable in cm_MakeDir that could result
2709 in crash if the requested directory could not be created.
2711 * Fix a the VolumeStatusNotification log messages to save
2712 the input string in the circular log buffer so garbarge
2713 won't be output to afsd.log when the log is dumped.
2716 * Increase the SMB Maximum Path length to 260 characters
2717 from 256 as that is what CIFS supports.
2719 * On 64-bit Windows, use a unique description string for
2720 the 32-bit AFS Shell Extension.
2724 * In the AFS Admin Library, replaced all of the
2725 Ubik_Call(XXX_RPC,...) calls with Ubik_XXX_RPC(...). This
2726 allows parameter checking and proper type conversions to be
2727 performed. This is crucial when passing parameters that are
2728 64-bit wide on systems in which 'long' is a 32-bit value.
2729 Otherwise, the parameter list on the stack is interpretted
2730 incorrectly and the function not only doesn't succeed but it
2731 might corrupt memory and if lucky crash the application.
2733 The only program that we care about that uses this library
2734 is the AFS Server Manager. Now it doesn't crash.
2736 * Fixed smb directory searches for non-wildcard names. The
2737 optimized function should not have set the bulk in progress
2738 flag. This would adversely affect later directory searches
2739 that required a bulk search.
2741 * Detect infinite recursions (up to 512 stat cache objects
2742 in a chain.) If an infinite recursion is detected, return
2743 an error to stop the CIFS client from querying us forever.
2745 * Executables executed out of AFS name space can have their
2746 pages swapped out by Windows. Those pages will be reloaded
2747 when they are required. However, if the cache manager no
2748 longer has a callback and all of the file servers with
2749 the volume data are down, then the application will crash
2750 due to the inability to read the required page.
2752 We now track the down time of each server. If the callback
2753 expires after the server that issued it went down and all
2754 servers are down, we delay the expiration to permit the data
2755 in the cache to be continue to be used by Windows or an
2758 * The Windows AFS client has always had very poor performance
2759 with regards to directory management. In the case of
2760 directory updates caused by create, rename,and delete operations,
2761 the AFS cache manager will now perform local directory updates
2762 when possible in order to avoid re-reading the directory from
2763 the file server. This will be done in whenever both of the
2764 following conditions are true:
2766 - the data version on the directory increased by one as a
2767 result of the operation
2769 - all of the directory buffers necessary to make the change
2770 are present in the cache
2772 * The error message in the AFS Control Panel describing the
2773 need to be a Windows Administrator has been replaced with
2774 a message describing the need to be in the AFS Admins Group.
2776 * If the vldb server list is empty when attempting to evaluate
2777 a cell mountpoint it was possible for a crash to occur. This
2778 can only happen if there is an existing mount point and the
2779 is a failure attempting to load the server list from the
2780 CellServDB file or via AFSDB queries.
2782 * dirty buffers only have their dirty bytes stored to the file
2783 server instead of the entire chunksize
2785 * the default chunksize has been increased from 128KB to 1MB
2790 * changed the enum values for cm_serverRef_t state info to use a
2791 private name space to avoid collisions (srv_)
2793 * added a srv_deleted state for cm_serverRef_t objects. This
2794 state is set when cm_FreeServerList() is called with the
2795 CM_FREESERVERLIST_DELETE flag set. cm_FreeServerList() may
2796 not always delete the cm_serverRef_t from the list if it is
2797 still in use by another thread. the srv_deleted state means
2798 the object's contents are no longer valid and it must be
2799 skipped. It will be deleted the next time the object is
2800 freed and the refcount hits zero.
2802 * the srv_deleted state is also used when a file server reports
2803 either VNOVOL or VMOVED instead of marking the cm_serverRef_t
2804 as offline. This is done to prevent additional usage of the
2805 stale vldb data while waiting for the update volume request
2808 * added a state field to the cm_volume_t object (enum volstate
2809 vl_ name space) that maintains the state of the volume based
2810 upon the states of all of the cm_serverRef_t and cm_server_t
2813 * modified cm_UpdateVolume() to set the state of the cm_volume_t
2814 RW, RO, and BK to either vl_alldown or vl_online. There can't
2815 be any other states because cm_UpdateVolume() destroys any
2816 previous knowledge we might have had regarding busy or offline
2819 * modified cm_UpdateVolume() to update the volume name in the
2820 cm_volume_t to the volume base name if the previous value was
2823 * modified cm_FollowMountPoint() to check to see if the volume
2824 name is a volume ID and if so call cm_GetVolumeByID instead
2825 of cm_GetVolumeByName. This ensures that volume IDs are always
2826 looked up as numeric values. There is no longer a need to
2827 maintain a separate cm_volume_t containing the string representation
2830 * Added a flags parameter to cm_GetVolumeByName() and cm_GetVolumeByID().
2831 The first flag is a "CREATE" flag which is set by all existing
2832 calls. The flag is not set by calls to cm_GetVolumeByID() from
2833 the server probe code when volume status is being updated. We
2834 do not want the server probe operation to result in additional
2835 turnover in the cached volume data. The second flag is NO_LRU_UPDATE
2836 which is set when the server probe code updates the volume status.
2837 This flag will be used to prevent the server probe operation from
2838 changing the order of the least recently used queue.
2840 * Modified cm_GetVolumeByName to ensure that only one cm_volume_t is
2841 allocated for a given set of normal, readonly, and backup volumes
2842 regardless of whether or not the volume is accessed via name or
2843 ID number. The cm_volume_t namep field is always the base name
2846 * Added a new volume state, vl_unknown. This state is used as
2847 the initial state for all cm_volume_t when the cache manager starts,
2848 for each cm_volume_t at creation, and for each cm_volume_t when
2849 recycling. The cache manager does not know the state of all
2850 volumes in the world, only those that are in the cache and for
2851 which it has queried the VLDB and hosting file servers.
2853 * modified cm_GetVolumeByName() to initialize the state of a
2854 volume to vl_unknown. The actual state will be set when a
2855 cm_VolumeUpdate() call completes successfully.
2857 * changed name of scache hash table variables to avoid ambiguity
2858 when adding hash tables for volumes
2860 * fix a buffer overrun in sys\pioctl_nt.c pioctl().
2862 * modified cm_UpdateVolume() to handle the case in which there is
2863 no RW volume but there is are RO volumes for a given base name.
2864 This is done by querying for the ".readonly" volume name if the
2865 base name does not exist in the VLDB. We never query for the
2866 .backup name because under the current usage model a .backup
2867 volume may only exist on the server that the read-write volume
2868 is located. If there is no RW volume, there can be no .backup.
2870 * Added four hash tables for cm_volume_t objects to improve the
2871 search time of cm_GetVolumeByID(), cm_GetVolumeByName() and
2872 cm_ForceUpdateVolume(). One each for Name, RWID, ROID, and
2873 BKID. Three ID hash tables are necessary as long as it is
2874 desireable to maintain a single cm_volume_t containing all
2875 of the related RW, RO, and BK volume data. Having the RW and
2876 RO volume data in the same object is necessary for the
2877 implementation of cm_GetROVolumeID() which returns either the
2878 RO or RW ID depending upon the existence of RO volume instances.
2880 * Added a volume LRU queue so that volume reuse becomes fairer.
2881 This does not replace the all Volumes list which is used when
2882 it is desireable to walk a list of all the volumes whose order
2883 is not going to change out from underneath you which makes it
2884 safe to drop the cm_volumeLock.
2886 * handles volume hash table updates where volume name to
2887 volume ID number changes. The volume name remains
2888 constant in the cm_volume_t. if a vos rename is performed,
2889 the name of the volume will change and the volume IDs will be
2890 updated. Subsequent access to the old volume ID will create a
2891 new cm_volume_t with the new name.
2893 * Added a daemon thread operation to query the state of volumes
2894 listed as busy or offline. cm_CheckBusyVolumes() calls
2895 RXAFS_GetVolumeStatus() for each volume ID that is marked vl_busy
2896 or vl_offline. If the volume is now online, the status on the
2897 volume is updated. The default period is 600 seconds. This can
2898 be configured with the BusyVolumeCheckInterval registry value.
2900 * Added prototype for smb_IoctlPrepareRead() which was missing a
2901 return type in the function definition.
2903 * Added volume id lists to the cm_server_t. These lists are
2904 allocated in blocks of ~32 IDs. When a cm_PingServer()
2905 detects a change in server state, the state of the cm_volume_t
2908 * Added volID to the cm_serverRef_t object. volID is used
2909 to identify the volume for which the object is a referral.
2910 cm_FreeServerList() uses the volID to remove the volume
2911 from the cm_server_t.
2913 * In cm_Analyze, when VNOVOL or VMOVED are received,
2914 call cm_ForceVolumeUpdate() to force a refresh of the volume
2917 * Added cm_VolumeStatusNotification() which is used at the moment
2918 to log volume status changes to the trace log. It will also
2919 be used as the access point to the File System Filter driver
2920 notification engine.
2922 * Added an all cm_scache_t list to cm_data. This replaces the use
2923 of the stat cache LRU queue when we need to enumerate all
2924 entries. The LRU list order is not static and when using it to
2925 enumerate all entries it can result in items being missed or
2926 items being processed more than once.
2928 * Modified cm_Analyze(). Instead of reseting the busy or offline
2929 state of a volume and forcing a retry of the operation
2930 cm_Analyze will defer to the background daemon thread that will
2931 update the state once every 600 seconds.
2933 * Added the automatic generation of a Freelance ".root" read-write
2934 mountpoint that refers to the root.afs volume of the workstation
2935 cellname at the time the mountpoint is created.
2937 * re-write cm_Analyze to make better use of the known volume
2938 status. VL_Server queries cannot result in CM_ERROR_ALLOFFLINE
2941 * renamed cm_CheckBusyVolumes to cm_CheckOfflineVolumes.
2942 busy volumes will be reset to srv_non_busy by the function
2943 but there is no mechanism for querying the busy state other
2944 than by attempting to access the resource.
2946 * cm_Analyze will query the state of an offline volume before
2947 deciding whether or not to retry when all volume instances
2950 * the algorithm used after an IP address list change detection
2951 has been modified. we now check the state of the VLDB servers,
2952 then reset the rx connections, and finally check the state of
2953 the file servers. Previously we flushed the rx connections
2954 and then checked the servers in whatever random order they
2955 were first discovered.
2957 * fix the volume status reported by "fs examine"
2959 * various improvements to the AFS Network Identity Manager Provider.
2960 - Don't access freed objects after the window has been destroyed.
2961 - Write back the AFS cell list in the identity configuration
2962 even when its empty.
2963 - Disable all use of the provider when AFS is disabled for a
2964 particular identity.
2966 * Fix cm_cell_t leak. When evaluating a mountpoint that uses an
2967 alias instead of the full cellname, double check the full cell
2968 name and see if we have a matching cell before allocating a new
2971 * Dump more data with "fs memdump". All volumes and scaches and
2974 * Reverse the order of the all cells list. This ensures that
2975 the workstation cell is always first in the list.
2977 * If we recognize the server from which a callback revoke was
2978 received, then we can ensure that we only invalidate the status
2979 information for cm_scache_t objects belonging to that cell.
2981 * When building for amd64, use APPVER 5.02 as we do not support
2982 64-bit Windows 2000 and explicitly state the machine type in
2985 * no longer treat VMWare network adapters as equivalent to a
2986 loopback adapter. For some unknown reason, publishing the "AFS"
2987 netbios name on them frequently results in a name collision
2990 * Use read locks whenever possible instead of write locks when
2991 walking the allCellsp list. This permits better use of multiple
2994 * In 1.5.20 a bug was introduced by the fake status when PRSFS_READ
2995 is not present on the directory optimization. It would report
2996 mount points and directory objects for which there was no
2997 status information as a zero length file. Instead we can use
2998 the vnode number of the object to determine if it should be
2999 considered to be a directory or a file. If it is odd it is a
3000 directory; even it is a file.
3002 * Give Back Callbacks to the file servers in response to suspend
3003 and shutdown power management events. Only give up callbacks
3004 to UP servers and do so with a very short RX dead timeout.
3007 * Fix a race condition in the SMB FID cleanup code that could
3008 result in a stat cache mutex never being released.
3010 * Fix the graphics files that are installed as part of the
3011 IBM Administrators Guide.
3013 * Change the behavior of pioctl path parsing to not follow
3014 symlinks or mount points during evaluation. This results in
3015 the FID being the symlink or mount point and not the object
3016 they point to. This is consistent with the behavior of the
3019 * Enable AFSDB record lookup for command line tools such as
3022 * CellServDB update 2007-07-05
3024 * Added a Volume Status Plug-in Interface that permits third parties
3025 to receive notifications when the volume state changes occur.
3027 * Ensure that all writes to the afsd_alloc.log file use CR-LF.
3029 * Prevent the afsprocmgmt.dll from being unloaded while the
3030 background thread it starts is running.
3032 * Avoid a race condition when closing smb_fid_t objects.
3035 * fix install path of IBM Quick Start Guide when installed by
3038 * fix the client so that it can write into directories with
3039 only 'li' permissions.
3041 * fix installation of 32-bit Shell Extension for 64-bit systems.
3042 Install it under its own HKCR keys.
3044 * modify the wix installer to remove and upgrade install packages
3045 with the same version number.
3047 * when flushing volume location data every two hours or in
3048 response to "fs checkvolumes", also flush the mount point
3049 links in order to force their re-evaluation.
3051 * aklog was copying an uninitialized buffer as part of the
3052 foreign cell automatic pts registration
3054 * when obtaining directory listings, check for read permission
3055 on the directory. If read permission does not exist, fake the
3056 status information for the directory entries. This improves
3057 performance, prevents the client from pounding the file server,
3058 and prevents leakage of cached status information.
3060 * "fs getcacheparms" output was incorrect.
3062 * NetIdMgr AFS Provider should only automatically add token
3063 acquisition for cell foo.com if the realm of the cell matches
3064 the realm of the identity.
3066 * Fix a null pointer dereference in cm_MergeStatus()
3070 * network identity manager provider:
3072 - Correct for the fact that pr_CreateUser tampers with the
3073 contents of the username input buffer.
3075 - Take into account that the krb4 credential data structure
3076 does not have separate realm fields for the user and service.
3077 When krb524 is used with a cross-realm ticket, the realm
3078 describes the service not the user.
3080 * Many integrated logon changes:
3082 - Fix eventlog reporting. Do not attempt to log an event
3083 if the event source registration fails. Use DebugEvent0
3084 instead of DebugEvent when there are no parameters.
3086 - Modify the LOOKUPKEYCHAIN macro to recognize ERROR_MORE_DATA
3089 - Fix the reading of Domain specific configuration for
3090 LogonScript and TheseCells. Previously the dwSize value was
3091 being overwritten so that subsequent RegQueryValueEx call
3094 - Fix a memory leak in the TheseCells reading code.
3096 - Add support for Domain specific "Realm" specification. The
3097 realm is the realm to be appended to the username. When
3098 logging in as a domain or to the local machine, the specified
3099 "Domain" name is not going to be a valid realm name.
3101 - Construct a proper principal name based upon the domain
3102 specified realm for use in obtaining tokens with KFW.
3104 - If the domain specified "TheseCells" list includes the
3105 default cell, do not obtain tokens twice.
3107 * An unprotected call to the Kerberos v5 function,
3108 krb5_free_context(), can result in an invalid memory access
3109 exception if the contents of the krb5 profile, krb5.ini, cannot
3110 be parsed correctly. The error is located within the afskfw
3111 library which is used by afscreds.exe, afssvrmgr.exe, and
3112 afslogon.dll. When the error is triggered within afslogon.dll
3113 a blue screen condition can occur when the winlogon.exe system
3114 process terminates unexpectedly.
3118 * Do not stop the afs client service with an assert
3119 if the cell name cannot be determined when responding
3120 to a "fs whereis" command.
3122 * Integrated Logon (afslogon.dll). No longer attempt
3123 to preserve Kerberos 5 tickets for the logon session
3124 as that job is now supported by KFW 3.1 and above.
3126 * The Network Identity Manager plug-in will attempt to
3127 remove expired service tickets from the cache.
3131 * CellServDB update - GCO Public CellServDB 09 Mar 2007
3133 * AFS Server Manager and AFS Account Manager data cache
3134 moved from HKCU\Software\TransarcCorporation to
3135 HKCU\Software\OpenAFS to prevent crashes due to data
3138 * Reduce the default UP server probe time from 10 minutes
3139 to 5 minutes to match UNIX. This will not solve the NAT
3140 port mapping timeout issue because most NATs default to
3141 a value less that 5 minutes and this value ensures that
3142 the afs client server will start probes 5 minutes after
3143 the last probe set ended. The probe period for any one
3144 server is therefore longer than 5 minutes. The more
3145 servers in the list the longer the time between probes.
3147 * Before marking a server as down in cm_Analyze, retry the
3148 request using a new rx connection to the server. If the
3149 cache manager is behind a NAT, it is possible that the
3150 response from the server is being sent to the original
3151 address:port from which the rx connection was established.
3152 This will appear to the cache manager as the server being
3153 down when in fact it is not. Forcing the use of a new rx
3154 connection will result in the server responding to the
3155 most recent address:port assigned to the client by the NAT.
3157 If the server is in fact down, it will simply take a little
3158 longer to notice it.
3162 * Removing a directory that contained entries was failing with
3163 the wrong error code being returned to the CIFS client.
3167 * rxdebug exponential back-off. don't take down a server that
3168 is already struggling to respond.
3170 * Network Identity Manager plug-in registers help on the Help
3171 menu if it is supported by NIM. (KFW 3.2 and above)
3173 * Plug-n-play networking did not always work after coming back
3174 from a Windows Suspend. The AFS client service failed to
3175 re-compute the LAN adapter to bind to and instead attempted
3176 to bind to all LAN adapters. This would polute the public
3177 network with the name "AFS" which could lead to conflicts
3180 * Build installers using Microsoft Hotfixes provided at
3181 http://tinyurl.com/243svk for Visual Studio C RunTime Library
3182 Daylight Savings Time issues.
3184 * "fs checkserver" does not require admin privileges unless
3185 the probe interval is being changed.
3189 * A reference leak was discovered in the mountpoint traversal
3190 functionality which resulted in cm_volume_t objects not becoming
3191 eligible for re-use. This will trigger a panic condition when
3192 more than the maximum number of volumes is accessed.
3194 * Fixed the success test for the "fs memdump" operation. Success
3195 is now reported as such. Added cm_volume_t object information
3198 * A data corruption error has been fixed. Write requests
3199 queued for background processing were not retried upon failure.
3200 This leads to inconsistencies between the contents of the file
3201 in the local cache and the contents of the file on the file
3204 The implemented fix is to re-queue any request that has failed
3205 due to timeout, volume busy, volume offline, server busy, or
3206 server offline. Requests are not processed while all servers
3207 for a volume are inaccessible. When servers become accessible
3208 the background request will be retried.
3210 In addition, if a background write fails, do not mark the buffers
3211 as clean. Failure to preserve the dirty flag prevents the buffers
3212 from being written to the file server in the future.
3214 This fix does not address the possibility that the credentials
3215 necessary to perform a write request might expire between the
3216 time the request is queued and when the request can be processed
3217 due to server accessibility.
3219 * The Freelance volume does not support per user ACLs. The
3220 cm_HaveAccessRights function did not handle this special case.
3221 Requests for privileges greater than read|lookup would result
3222 in afsd_service.exe entering an infinite loop. ISOBuster is
3223 known to cause this behavior.
3225 * When deleting files, the SMB Server would attempt to obtain
3226 the access permissions for the parent directory instead of
3227 for the file itself. This would result in an incorrect
3228 assessment of whether or not the file can be deleted.
3230 * The Network Identity Manager AFS plug-in has had its string
3231 tables updated for consistency with the language used by NIM.
3232 All credentials are referred to as credentials instead of tokens.
3233 Drop-down combo boxes now in fact drop down.
3235 * The AFS Salvager was not properly being built on Windows.
3236 This is relevant only if the AFS servers are in use.
3238 * The default signal handlers for the AFS servers incorrectly
3239 processed SIGQUIT. Instead of stopping the server, an
3240 exception was raised.
3242 * Fix another volume refcount leak. This one was triggered by
3243 an inability to contact the vlservers in order to update the
3246 * Fix checklist control used by the drive mappings tabs in
3247 afscreds.exe and afs_control.exe.
3249 * Improve output of "fs memdump" related to buffer management.
3251 * Fix refcount leak of cm_buf_t objects when prefetching buffers
3252 already in the process of being fetched.
3254 * Improve cache performance by computing the buffer hash table
3255 size based upon the number of buffers so that the average
3256 number of entries in a bucket is seven.
3258 * Improve cache performance of read/write data by locally
3259 updating the dataVersion of all buffers associated with the
3260 file on each StoreData or StoreStatus operation provided that
3261 the dataVersion was incremented by one and the dataVersion of
3262 the buffer matched the original dataVersion of the cached
3263 status information for the file.
3265 * When recycling cm_scache_t objects, make sure that we
3266 clear all of the previous fields. It is no longer the
3267 case that we can assume that the stat cache object is being
3268 obtained only after a successful FetchStatus call.
3269 InlineBulkStatus will allocate stat cache objects for which
3270 the no status information is available.
3272 * When a file is being opened with the OPEN_ALWAYS disposition,
3273 do not require write permissions when the file already exists
3274 unless of course the file is being opened requesting write
3277 * When a file has been opened for read-only access, if the
3278 application requests an exclusive lock, obtain a read-lock
3279 instead of a write-lock. The purpose of the exclusive lock
3280 in this situation is to prevent changes to the file and the
3281 full file read lock will provide this guarrantee.
3286 * Enforce Unix Mode Write bit to prevent file deletion of read only
3289 * The CIFS READ and LIST permissions use the same value 0x0001.
3290 When a directory is being accessed, check for PRSFS_LOOKUP
3291 instead of PRSFS_READ.
3295 * Fix computations of short names of mount points. They were
3296 broken as part of the directory search optimization. (1.5.13a)
3298 * When all buffers are in use, periodically sleep the threads
3299 waiting for a buffer to become free while the global buffer
3300 lock is not being held in order to permit those threads that
3301 can free buffers to do so.
3303 * Power Management improvements. Maintain a global flag that
3304 specifies whether or not the service is in a suspend state.
3305 Do not panic if Netbios() returns NRC_BRIDGE meaning that the
3306 lana is no longer valid. Instead, stop the listener threads
3307 and if all listener threads are stopped, reset the lana_list.
3308 Allow the cm_Daemon() thread to periodically check the state
3309 of the smb listeners. If they are all stopped and the service
3310 is not suspended, attempt to restart them. If there are no valid
3311 lanas, return to the stopped state.
3313 * CreateX and NTCreateX use cm_CheckNTOpen() to test whether or not
3314 the user is permitted to obtain read or write locks. This function
3315 would obtain the lock and then drop it returning whether or not
3316 the lock could be obtained. If the lock was in fact required,
3317 CreateX/NTCreateX would then obtain it with cm_Lock(). The problem
3318 of course being that this pattern results in three RPCs to the
3319 file server (lock, unlock, lock). This is reduced to one RPC by
3320 implementing cm_CheckNTOpenDone() which frees the allocated byte
3321 range lock from cm_CheckNTOpen() after the cm_Lock() call is
3324 * Optimize the SMB FindFirst processing for the case in which there
3325 are no wildcards. If a cm_Lookup succeeds, generate a response
3326 from the stat cache entry we have a callback for. Its cheaper
3327 than walking each buffer of directory data associated with the
3328 directory stat cache entry when we have thousands of directory
3331 * Increase the SMB Ioctl MaxData size to support larger token sizes
3333 * Disable autorun of the AFS Server Wizard when the AFS Server is
3336 * Add AFS Server service dependencies on tcpip and PNP_TDI
3338 * Remove the AFS Server Configuration Tool shortcut
3340 * Add Freelance mode awareness to AFS Server Configuration Tool
3344 * Unlike previous versions of the OS, Vista performs a shutdown on
3345 the Microsoft Loopback adapter just like it would on a real adapter.
3346 This causes the smb_Listener threads to trigger a panic during a
3347 suspend/hibernate power event.
3349 The fix is to unbind from the network adapters in response to a
3350 suspend/hibernate power event and then rebind to the adapters
3351 when a resume power event is received. Note that the resume events
3352 are not reliably delivered so it is possible the afs service will
3353 not be accessible. However, this is the best we can do.
3355 * don't allow environment variables we set in afslogon.dll to be
3356 inherited by child processes
3358 * force an rx connection reset if VICECONNBAD or VICETOKENDEAD are
3361 * The old dirty buffer synchronization algorithm had a buf_IncrSyncer
3362 thread walking the all buffer list periodically searching for dirty
3363 buffers to write to the file server. This had several negative
3364 results. The algorithm ate up ever increasing amounts of CPU time
3365 even when AFS is idle as the size of the cache increases. Also,
3366 buffers were written to the file server in an order based upon the
3367 original buffer allocation which has nothing to do with the order
3368 in which the buffers became dirty.
3370 The new algorithm maintains a dirty buffer list. Items are added
3371 when the buffer is originally marked dirty. A buffer is only
3372 removed from the list by the buf_IncrSyncer when the buffer is no
3373 longer dirty. If the list is empty the thread goes back to sleep
3374 immediately without additional processing requirements.
3378 * Enable renaming of directories from the Explorer Shell when
3379 Offline Folders is enabled
3381 * An error is reported in response to the SMB ReceiveCoreFlush
3382 request when an error has not occurred.
3386 * optimize dirty buffer flushes in response to suspend operations
3387 do not flush if only loopback adapter is installed
3388 do no retry dirty buffer writes
3390 * do not reset netbios when performing loopback checks. doing
3391 so will break the existing lana bind
3393 * remove a deadlock in smb3.c QPathInfo
3397 * Add NetIDMgr AFS Plug-in
3399 * "Show Tray Icon" checkbox removed from afs_config.exe
3401 * Disable DNS Registrations for the Loopback Adapter and make sure
3402 that Netbios is turned on.
3404 * A reference count leak on stat cache objects was fixed
3406 * For debugging, tread ids of writers are now tracked on locks
3408 * Corrected an incorrect lock that was being held but not
3409 freed in a rare error case
3411 * Changed trace log end of line to CR-LF so it can be read
3414 * modify the buffer synchronization thread to write all dirty
3415 buffers and sleep only when there is no work to be done
3417 * improve behavior of cache manager when large numbers of
3418 temporary files are being created by multiple clients within
3421 * do not permit dirty buffers that belong to deleted files
3422 to be written to the file server. its a waste of bandwidth
3424 * do not leak stat cache references when the file server
3427 * afscreds: do not permit an attempt to obtain tokens if the
3428 afs service could not be started. this was causing stack
3429 overflows on Vista after resuming from a suspend
3433 * better handling of dot directory
3435 * fs sq and fs mkm now behave as on UNIX
3437 * add license to installers
3439 * Return Invalid Parameter error for unknown InfoLevels in
3440 SMB Tran2 Search Directory.
3442 * Correctly use WSAStartup and WSACleanup within the RX library
3444 * Check that afsd_service.exe is running before performing
3447 * Force AFS Shell Extension to load resource DLL using
3448 afs_shl_ext.dll as a base name.
3450 * (Wix) install afs_shl_ext_XXXX.dll and afs_creds_XXXX.dll in
3451 OpenAFS\Client\Program
3453 * Fix SMB_SET_FILE_END_OF_FILE_LENGTH
3455 * Do not panic if the maximum number of volume entries are in use
3456 and one of them can be recycled.
3460 * Fix the code that reads the backconnnectionhostnames value
3461 so that there is not an uninitialized variable
3465 * integrated logon could cause winlogin.exe to blue screen windows
3466 if it was called with a domain that does not exist on the machine.
3467 this is possible during a sysprep.
3469 * corrections to cm_buf_t handling to ensure that buffers are
3470 removed from queues before they are freed and that holds are
3471 obtained on the next buffer in a chain before releasing the
3474 * do not generate MessageBox dialogs on network errors. they
3475 cause more trouble then they are worth.
3477 * move GlobalAutoMapper into a separate thread so that it does
3478 not prevent service startup if the WNetAddConnection2 call
3481 * restrict hard dead and connection timeouts to the CIFS Session
3482 Timeout value. This is being done to try to prevent virtual
3483 circuits from being destroyed prematurely.
3485 * Fix the code that writes the backconnectionhostnames value
3486 to ensure that the data buffer is written with the correct
3489 (there were no windows specific changes between 1.5.4 and 1.5.5)
3493 * Corrections to the Integrated Logon functionality that permits
3494 Kerberos 5 TGTs to be communicated into the user session. The
3495 transfer is now more secure and succeeds when the user does not
3496 have read permission to the %SystemRoot%\TEMP directory.
3498 * NSIS installer updated to version 2.18
3500 * WiX installer updated to version 2.0.4221.0
3502 * Updates to the IFS functionality for 64-bit correctness and
3503 permit compilation. The IFS functionality works for pioctl()
3504 operations but not much else at this point.
3506 * A major re-write of the OpenAFS implementation of the CIFS Tran2
3507 query and set functions.
3509 * Ignore the Windows CIFS client session timeout entirely. Timeout
3510 only after the AFS Hard Dead Timeout period expires.
3514 * move headers, libraries, and samples from OpenAFS\Client to
3517 * remove the auto-registration of AFS IDs for foreign cells
3518 from integrated logon. this permits afslogon.dll to avoid
3519 initializing the RX library. afslogon.dll is a network
3520 provider library and as such it is loaded into all applications
3521 which perform a CreateProcessAsUser call. Initializing the
3522 RX library spawns background threads to listen for inbound
3523 packets. When the network provider is unloaded from the
3524 process, the RX library does not cleanup the threads. If they
3525 continue to execute after the library is unloaded, the process
3528 * implement inline bulk status calls. this will significantly
3529 improve the performance of directory listings.
3531 * prevent a crash when evaluating mountpoints to volumes that do not
3534 * new command: fs uuid [-generate]
3535 displays the current uuid and allows a new one to be generated
3536 on the fly without restarting the afs client service.
3538 * fix the 64-bit installers to properly document the install
3541 * move help files to the proper location so they can be located by
3542 the system tray tool and the control panel.
3544 * "over quota" and "disk full" errors were not being reported to the
3547 * largefile support has been added. Files greater than 31-bits can no
3548 longer be written to file servers which do not support large files.
3549 When file servers support large files, files up to 64-bits can be
3552 * cm_SyncOp could return with the input buffer unlocked upon error
3553 which would result in a panic condition. This bug was introduced
3556 * vlserver failover when freelance mode did not work. This prevented
3557 the root.afs volume from being loaded if the first vlserver could
3562 * A bug in the CIFS Standard Info response packet caused directories
3563 to appear as standard files. This prevented Microsoft Vista from
3564 being able to walk the AFS directory tree.
3566 * Mount points must be reported as directories in response to
3567 Query Path Info operations.
3569 * A deadlock was detected when performing "fs flushall" if the
3570 file server reports VNOVNODE. The scp->createBufferLock is
3571 already held by the current thread and there is no mechanism
3572 to propagate the knowledge. Therefore, an alternate mechanism
3573 for clearing the cache must be developed. A new function
3574 cm_RecycleSCache(scp,flags) has been extracted from
3575 cm_GetNewSCache(). This function performs the task of recycling
3576 an cm_scache_t object. When called from cm_GetNewSCache() with
3577 no flags the expectation is that there are no associated buffers
3578 that are queued to be read or written. When called from
3579 cm_Analyze() with the CM_SCACHE_RECYCLEFLAG_DESTROY_BUFFERS
3580 flag, any queued buffers will be de-queued and marked as if
3581 the operations were performed so the data can be discarded.
3583 * CM_ERROR codes were logged as "unknown" by cm_Analyze.
3585 * In response to VNOVNODE, the parent is only discarded if the current
3586 cm_scache_t is not a directory.
3588 * In the Ioctl FlushFile and FlushVolume functions, there were no
3589 checks to protect against flushing the Freelance SCache entries.
3591 * In FlushFile, the wrong cm_scache_t object was being released.
3593 * In cm_GetNewSCache, do not allow Freelance SCache entries to be
3594 recycled. Choose a new entry if cm_RecycleSCache fails.
3596 * Speed up the performance of the cache manager by not
3597 holding the smb_fid_t mutex across calls to cm_SyncOp
3600 * Ensure that all smb_fid_t flag references are protected.
3602 * Remove a deadlock between smb_fid_t mutex and
3603 smb_rctLock introduced in 1.4.1.
3605 * Fix aklog AFSID lookup to use not send the realm
3606 name when the realm is the local realm for the
3607 cell. The PT Server doesn't strip the realm
3608 name when it is the local realm.
3610 * Treat "\\afs\*." as an alias for "\\afs\all"
3612 * Dynamically adjust the priority of server threads
3613 based upon the age of the CIFS request that is
3614 being processed. The priority is bumped one level
3615 for every 15 seconds of processing time.
3617 * The original openafs contribution mis-used the
3618 cm_GetCallback function. This function should only
3619 be called as a side effect of a call to
3620 cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK). The way it
3621 was being called results in a FetchStatus RPC being
3622 performed even when the client already has a valid
3623 callback and does so in a manner that destroys the
3624 synchronization of the threads that are calling
3625 cm_SyncOp. If nothing else this results in a
3626 significant performance penalty.
3628 * More changes to cleanup of smb_vc_t objects in
3629 order to prevent race conditions.
3631 * Reduce the amount of Power Management event logging
3634 * If the client detects that the IP address has
3635 changed, force the use of new RX connections. This
3636 enables the file server to recognize that the client
3639 * Remove a race condition and deadlock associated
3640 with cleanup of smb_vc_t objects.
3642 * Fix the ANSI filename support. One function that
3643 required it was missing the translations.
3645 Changes not present in 1.4.1 and since 1.5.0: 1.5.1 released
3648 * Add a DllMain() function to afspthread.dll. Perform
3649 cleanup of thread local storage, rx queues, critical
3650 sections, event handles, etc. This was implemented
3651 to avoid crashes in afspthread.dll when applications
3652 which load/unload the dll attempt to terminate.
3654 * Lock request behavior has been altered. As of this
3655 release a locally allocated lock will be used whenever
3656 the volume is RO or the requesting user's permissions
3657 are 'rl'. This will allow end users to execute programs
3658 from RW volumes when they do not have the 'k' privilege.
3660 Note that a recent discovery indicates that the 'k' bit
3661 is only supposed to control the ability to obtain read
3662 locks. Write locks are to be controlled by the 'w' and
3665 * Code has been added to enable the cache manager to be
3666 able to obtain the AFSID of the authenticated user.
3667 This is required to be able to determine when "Insert"
3668 privileges should apply. AFS considers a file to be
3669 a new file whenever the owner's AFSID matches the
3670 AFSID of the user. The current implementation relies
3671 on the name specified during the SetToken call to be
3672 correct. This will be replaced in the future with a
3673 new RPC that allows the AFSID of the authenticated
3674 user to be obtained.
3676 * A 32-bit-tools installer for use on 64-bit systems
3677 is now available. These tools are for use with 32-bit
3678 versions of Kerberos for Windows or JPSoftware's 4NT.
3680 Since 1.4.1 rc10: 1.4.1 final released (9 April 2006)
3682 * Remove the use of the "active_vcp" variable. This variable
3683 no longer serves any useful purpose and was the cause of
3684 race conditions that produced under counts of the smb_vc_t
3687 * Change default minidump output to include data segments
3689 * restore access to cm_IoctlGetVolumeStatus and
3690 cm_IoctlSetVolumeStatus calls for AFS pioctl operations.
3691 Still need to figure out why they are being called
3692 for SMB/CIFS IOCTL operations instead of Trans2GetFileInfo.
3694 Changes not present in 1.4.1: 1.5.0 released (16 Feb 2006)
3696 * Update build system to include the "sys_name" in the
3699 * optimize the number of calls to obtain the smb_user_t for a
3700 smb virtual circuit.
3702 * add logging of server addresses, status and capabilities when
3705 * prevent multiple threads from pinging the same server simultaneously.
3706 instead, threads block and then use the result of the ping that was
3707 already taking place.
3709 * Add registry values to allow the defaults for daemon checks to be altered
3710 at startup. Log values to afsd_init.log.
3712 daemonCheckDownInterval
3713 daemonCheckUpInterval
3714 daemonCheckVolInterval
3715 daemonCheckCBInterval
3716 daemonCheckLockInterval
3717 daemonCheckTokenInterval
3719 * Add EventLog messages for transitions in fs crypt state.
3721 * Byte range locking (when locks will be obtained from the
3722 is still under discussion)
3724 * 64-bit Windows operating systems now supported
3725 (XP SP2 64-bit, 2003 SP1 64-bit, 2003 R2 64-bit)
3726 This is a native 64-bit implementation of OpenAFS for Windows.
3727 There is currently no 64-bit version of KFW and therefore
3728 there is no (easy) method for obtaining AFS credentials.
3729 32-bit KFW 3.0 can be used with 32-bit OAFW libraries
3730 to obtain AFS tokens but there is not for the feint of
3733 * A faster implementation of the fcrypt algorithms improves
3736 * The afsd_service.exe now supplies Event Log Message Text.
3737 The event log no longer generates warning text due to lack
3738 of message configuration.
3740 * When afs servers are unreachable and network stack returns
3741 a WSAEHOSTUNREACHABLE error, the AFS client will immediately
3742 mark the attempted server as being down. This allows for
3743 very fast response to the calling application. This change
3744 has a pronounced improvement when the Windows Explorer is
3745 used on disconnected mobile computers.
3747 * Added DebugOutputString debugging of the RX library
3749 Since 1.4.1 rc9: 1.4.1 rc10 released (10 March 2006)
3751 * Fix the rx-lwp implementation to randomly generate rx call
3752 identifiers which are supposed to be unique at a given point
3753 in time. This affects some of the command line tools used
3754 for debugging but not the AFS Client Service itself which
3755 uses the rx-pthreads implementation.
3757 * xstats_fs_test.exe and xstats_cm_test.exe are now built for
3758 Windows although they are not distributed
3760 * don't generate an exception when logging unexpected SMB packets
3762 * the data structures used by cm_IoctlGetVolumeStatus and
3763 cm_IoctlSetVolumeStatus are not compatible with the CIFS
3764 Get/Set File Information calls so don't bother executing them.
3768 * Fix data written to registry as part of BackConnectionHostnames
3771 * Fix initialization bug when allocating cm_server_t objects that
3772 would cause servers to appear to be down after each allocation
3773 or renewal. Server would appear up after the next check servers
3778 * Windows will return WSAECONNRESET to the next receive packet
3779 operation on a socket if an ICMP packet is received in response
3780 to a packet that was sent and could not be delivered because
3781 the port was unreachable. We have to treat this as
3782 WSAEWOULDBLOCK so that we can retry the select() check and
3783 not mark the connection as bad.
3785 Since 1.4.1 rc6: 1.4.1 rc7 released (14 Feb 2006)
3787 * The client manager maintains a least recently used list of
3788 all the ACLs. If an ACL is already the most recently used
3789 do not bother to remove it from the list and place it on the
3792 * As part of the process of adding a new server, set the
3793 server is down flag and then perform a ping. If the server
3794 is up, the flag will be reset. If the server is down, the
3795 client will not timeout unnecessarily while attempting to
3798 * If a Mountpoint String is the empty string, return PATH NOT
3801 * Modify behavior of the Network Provider to avoid querying the
3802 profile location if integrated logon is known to be disabled.
3804 * Correct behavior of 'LogoffPreserveTokens'. When set, tokens
3805 will always be preserved regardless of whether or not integrated
3806 logon was used to obtain tokens. When unset, tokens will be
3807 destroyed regardless of whether integrated logon was used to
3808 obtain the tokens unless the profile is stored in AFS.
3810 * Disable LogoffTransferToken functionality. The SMB V3 Logoff
3811 message is being delivered even when the user is not logging
3812 off of the computer.
3814 * Fix a deadlock condition introduced in rc6.
3816 * Cell names are case-insensitive. Since the founding of OpenAFS
3817 there has been a bug that treated "cell", "Cell", and "CELL" as
3818 different cells. This would result in what appeared to be
3819 random disappearance of user tokens or a failure to authenticate
3820 even when tokens were listed by the "tokens" command.
3822 Since 1.4.1 rc5: 1.4.1 rc6 released (2 Feb 2006)
3824 * Additional auditing of the smb code revealed that smb ioctl
3825 operations were failing to free smb_fid_t objects
3827 * The File Identifiers assigned to smb_fid_t objects could be
3828 given the value 65535 which is 0xFFFF which is also the
3829 INVALID_HANDLE value. Both 0 and 65535 are now treated as
3832 * Audited all assignments to fields and ensured that they are
3833 protected by obtaining the relevant mutex object.
3835 * Re-wrote the VNOVNODE handling to avoid a potential deadlock
3836 if both the current cm_scache_t object and its parent are
3837 both returning VNOVNODE.
3839 * Removed all the logic associated with the 'dead_vcp' variable
3840 which must have been added for debugging because its used
3841 was never thread safe and it truly served no purpose.
3843 Since 1.4.1 rc4: 1.4.1 rc5 released (25 Jan 2006)
3845 * Re-enabled LogoffTokenTransfer and LogoffTokenTransferTimeout
3846 registry settings. Default is ON and 120 seconds. If set to
3847 OFF, tokens will not be destroyed.
3849 * Fixed a number of race conditions and reference count problems
3850 associated with global queues.
3852 * Added a method to detect abandoned smb virtual circuits so their
3853 associated file handles, locks, user tokens, and smb session
3854 resources can be released.
3856 * Added a mechanism to allow user tokens to be preserved
3857 during integrated login and be freed after logoff.
3859 * Check access permissions to the file before the file is
3862 * More improvements to the handling of SMB Virtual Circuits
3863 when SMB sessions are closed.
3865 * Correct a race condition introduced in rc4 that could result
3866 in a crash of afsd_service.exe.
3868 Since 1.4.1 rc3: 1.4.1 rc4 released (11 Jan 2006)
3870 * Properly handle pre-mature SMB session closures.
3871 File handles, tree connections, and user sessions are now
3872 freed when the virtual circuit is closed.
3874 * smb_username_t and smb_user_t objects are now properly reference
3875 counted and freed as appropriate.
3877 * disable to use of AFS file locks. byte range locking is still
3878 implemented but all locks are managed locally. This prevents
3879 multiple processes from editing the same file on the same
3880 machine but it does prevent multiple processes on separate
3881 machines from modifying the file.
3883 Since 1.4.1 rc2: 1.4.1 RC3 released (4 Jan 2006)
3885 * A fix to "fs setserverprefs -vlserver". Multiple calls with
3886 the same server parameter would result in a panic condition
3887 if the server had not already been contacted by the AFS
3890 * Better warnings are written to the afsd.log file when
3891 the Windows RPC Protocols are not properly configured as
3892 is often the case on Windows 2000.
3894 * fix the procmgmt library to restore C RTL signal handlers when
3895 it is being detached from the process prior to process termination.
3897 Since 1.4.1 rc1: 1.4.1 RC2 released (30 Nov 2005)
3899 * close a lock leak in the rx library that could be triggered while
3902 * prevent removal of drive mappings in response to afscreds.exe -M
3904 * add logic to process VNOVNODE in cm_Analyze. Force re-evaluation of symlink
3905 strings and flush the stat cache entry.
3907 * prevent server objects from being freed if user preferences are set.
3909 * Update WIX support to 2.0.2419.0
3911 * Release all locks on the server at service shutdown.
3913 * A down server should be marked up as long as it responds to the client.
3914 It doesn't matter if the server returned an error as long as the error
3915 is something other than VRESTARTING. If there is such an error, all
3916 connections on that server should be marked for replacement.
3918 * Change the check "up" servers period from once per hour to once every
3919 ten minutes. This matches the behavior of the Unix cache manager.
3921 Post 1.4.0 revisions: 1.4.1 RC1 released (21 Oct 2005)
3922 * Byte Range Locking support has been added. Architectural documentation
3923 is located within src/WINNT/afsd/cm_vnodeops.c. Best efforts are
3924 made to enforce Windows mandatory locking model using locally managed
3925 byte range locks with full file locks obtained from the file server.
3926 Read-only volumes provide the user with an implied locking privilege.
3927 Read-write volumes require that a user be granted the locking privilege
3928 in order for a lock to be obtained. As most Windows applications use
3929 Shared file access modes, they require the ability to obtain locks in
3930 order to access files.
3932 * Implement error translation functions to convert Universal Error Codes
3933 output by modern servers to local errno and winsock error values.
3934 This fixes the error message problems using pioctl() calls from fs.exe
3935 and the AFS Shell Extension.
3937 Since 1.4.0 rc8: 1.4.0 final tagged (20 Oct 2005); released (31 Oct 2005)
3938 * Allow the AFS Client Service to set the firewall configuration by
3939 retrying until after Windows determines the boot period is over.
3941 * Do not attempt to configure the firewall on Windows 2000, XP pre-SP2,
3944 * Detect IP address changes from within the AFS Client Service for the
3945 purpose of checking down servers sooner
3947 * If the file server or vldb server report RXKADUNKNOWNKEY, return
3948 SEC_E_NO_KERB_KEY to Windows. This provides an indication to the
3949 user that there is a Kerberos key problem. Previously the user would
3950 receive no feedback when there was something wrong.
3952 * Fix the large integer library support to properly handle values greater
3955 Since 1.4.0 rc7: 1.4.0 rc8 released (15 Oct 2005)
3956 * Updated CellServDB
3958 * Fixed file name pattern matching. Question marks can match nothing
3959 if found at the end of a component.
3961 Since 1.4.0 rc6: 1.4.0 rc7 released (8 Oct 2005)
3962 * Kerberos 5 ticket lifetimes should adhere to KFW Leash lifetimes
3964 * The MSI installer now has a new Product ID for each build. This
3965 is to allow for automated upgrades. The Upgrade ID is remaining
3968 * Volumes that were cached and then subsequently were salvaged,
3969 moved or otherwise taken temporarily offline (as opposed to being
3970 busy) would become unreachable. A distinction is now made between
3971 a volume being offline and a server being down.
3973 Since 1.4.0 rc5: 1.4.0 rc6 released (2 Oct 2005)
3975 * Fixed several AFS Shell Extension issues:
3976 - Checking Server Status would only work if a specific cell was
3978 - If servers were down, the Server Down dialog would write to
3979 invalid memory and cause explorer.exe to crash
3980 - Default Push Buttons were not specified for many dialogs
3981 - The extension would be unloaded by the Explorer Shell and
3982 never be reloaded under some circumstances.
3984 * Fixed the AFS System Tray menu to disappear if another Window
3985 was selected before a menu item was selected
3987 * Added volume owner and group information to "fs examine" output
3989 * Added fs and registry support for enabling/disabling/clearing
3990 rx statistics gathering.
3992 Since 1.4.0 rc4: 1.4.0 rc5 released (25 Sep 2005)
3994 * New Product Code GUIDs issues for 1.4 and 1.5 branches. Separate
3995 GUIDs for each platform.
3997 * Documentation directory structure and default installation rules
4000 * New OpenAFS for Windows HTML Release Notes installed
4001 as part of the documentation set.
4003 * Add thread id to rx debugging messages
4005 * When all servers are offline, return a bad network path error
4006 immediately. This is necessary to allow Explorer to do the
4007 right thing and not hang. The background daemon thread will
4008 check the down servers every 30 seconds to see if they are
4011 Since 1.4.0 rc3: 1.4.0 rc4 released (14 Sep 2005)
4013 * The AFSCache file has been modified to store both the
4014 serial number of the volume on which the file resides
4015 and the SID of the local machine. If either change,
4016 a new UUID is generated.
4018 * add "fs flushall" command that forces all data buffers
4021 Since 1.4.0 rc2: 1.4.0 rc3 released (4 Sep 2005)
4023 * add support for '/' instead of '\' in pioctl() calls
4025 * Apply AFS Client Admins group protect to AFS Shell Extension
4027 * Add support for \\afs\<name> to most AFS Shell Extension
4028 commands except the symlink methods.
4030 * when installing the Microsoft Loopback Adapter, enable
4031 MS Client for Networks and prevent an install failure
4032 by not calling CoInitialize twice in the same thread.
4034 * reload cell vldb values from the CellServDB every two
4035 hours in case it changes
4037 * When updating cell information from DNS, be sure to set
4040 * Add support to allow use of \\AFS\<foo> where <foo> is
4041 either a mount point or symlink. As <foo> is normally
4042 treated as a share name, we transform it into \\AFS\all\<foo>
4045 * Init 'code' to prevent false errors when integrated logon
4046 is disabled and the service is not running
4048 Since 1.4.0 rc1: 1.4.0 rc2 released (28 Aug 2005)
4050 * Removed trace log messages that were hampering performance.
4052 * Fixed a deadlock that was being triggered by editing Word
4053 documents stored within AFS with WinWord 2003.
4055 * Bit 3 of the TraceOption registry value is now used to set the
4056 default for "fs trace" in the non-Debug builds. The new default
4057 is off for release builds and on for debug builds.
4059 * Bit 2 of the TraceOption registry value can be used to turn on
4060 real-time output of debug log entries to the Windows Debug
4061 Output monitor interface. This data can be viewed with tools
4062 such as Sysinternal's DbgView.
4064 Since 1.3.87: 1.4.0 rc1 released (17 Aug 2005)
4065 * When the cache manager reports ALLBUSY or ALLOFFLINE for
4066 a resource we should not be returning status codes such
4067 as Network Busy or Remote Host Down because these errors
4068 will imply that the AFS Client Service is Busy or Down
4069 and that is simply not the case. Instead we will return
4070 Bad Network Path as the path is temporarily not available.
4072 Instead of returning Sharing Paused when there is a
4073 timeout error, return Timeout. Once again, the AFS Client
4074 Service is not paused.
4076 * afscreds.exe would display an Obtain Creds dialog when
4077 the expired credentials reminder was triggered even
4078 if there was no network path to the KDC. This is prevented
4079 by adding KDC probe logic to the reminder thread.
4081 * afscreds.exe would display expired tokens no differently
4082 than unexpired ones. This would make it difficult for the
4083 user to distiguish when the tokens were expired. For the
4084 English build added a new resource string "(expired) that
4085 is displayed instead of the expiration time.
4088 * "fs wscell" when executed in freelance mode will return the
4089 name of the cell configured in the registry. The root.afs
4090 volume is not loaded from this cell, but it is used for the
4091 default for aklog and integrated login.
4093 * "fs mkmount, fs rmmount, symlink make, symlink remove" will
4094 no longer work on \\AFS\all when freelance is being used
4095 unless the user is a member of the "AFS Client Admins" Windows
4098 * some more checks performed during persistent cache validation.
4101 * Optimize calls to cm_CheckServer from cm_Analyze. Only check
4102 down servers for the current cell not all cells.
4104 * If the scache LRU list becomes corrupted, the AFS Client Service
4105 might enter an infinite loop at startup. Detect the corruption
4106 and discard the cache contents in this case.
4108 * Fixed the Explorer Shell's remove mountpoint functionality.
4110 * Fixed a deadlock caused by the holder of cm_aclLock attempting to
4111 obtain a mutex lock on a cm_scache_t object whose lock is held by
4112 a thread attempting to obtain the cm_aclLock.
4115 * Really fixed DNS AFSDB queries to ensure that "csail" cannot be
4116 misinterpretted as "csail.mit.edu" when the resolver
4117 is configured to append ".mit.edu" to failed DNS queries.
4119 * Added a new registry key, "LogoffPreserveTokens" (see registry.txt),
4120 that can be used to force the preservation of user tokens upon logout.
4122 * Update the NSIS install scripts to use NSIS 2.07. This release adds
4123 recommended installation categories: AFS Client, AFS Administrator,
4124 AFS Server, AFS Development Kit. Each category includes a different
4125 default subset of the OpenAFS components.
4127 The OpenAFS logo is now associated with the NSIS uninstall entry
4128 in the Add/Remove Programs control panel.
4130 * The user name associated with AFS tokens when obtained with
4131 integrated login, the afs systray tool, or aklog will always
4132 include the full Kerberos 5 user principal name regardless of
4133 whether or not the cell is local to the realm.
4135 * Modify integrated login so that it does not enter an infinite
4136 loop if the service is not set to auto start.
4138 * Added asetkey.exe used to set a Kerberos 5 key for use by the
4141 * Added uninstall.exe to Wix installation
4143 * More modifications to algorithms used to wake sleeping threads.
4147 * Changes to the algorithms used to wake threads when they are
4148 put to sleep because of conflicts over buffer operations.
4151 * The OpenAFS integrated logon obtains Kerberos 5 tickets as part of
4152 the process of producing AFS tokens. The tokens are stored within
4153 the AFS Client Service but the Kerberos 5 tickets have been discarded.
4154 New functionality has been added to temporarily copy the tickets to
4155 a file ccache during the short window between the NPLogonNotify and
4156 WinLogon Logon Event. The Logon Event handler starts a process as
4157 user, afscpcc.exe, that copies the tickets into the default ccache.
4158 The file ccache is then destroyed by the event handler.
4160 * During synchronization operations on stat cache entries, if threads
4161 are forced to sleep it was possible that the threads would not be
4162 woken under heavy load.
4164 * Enforce the delete on close flag specified by the NTCreateX()
4167 * A race condition in the rx library was fixed that resulted in
4168 a deadlock in rx_NewCall when the RX_CONN_MAKECALL_WAITING flag
4169 was cleared when threads were still waiting to use the
4170 connection. Also, fixed a potential case in which the
4171 connection could be destroyed while threads were waiting to use it.
4173 * The DNLC freelist has been seen to become corrupted with still
4174 active entries being stored on the freelist. Changes were made
4175 to perform a better job of cleaning entries before placing them
4176 onto the freelist; marking them with the correct magic value when
4177 purging the list; and allowing the list to be purged on startup
4178 if the cache validation fails.
4180 * Windows Crash Reporting does not get triggered for afsd_service.exe
4181 because it provides its own top level exception handler. This patch
4182 places an explicit call to ReportCrash() on platforms which support
4185 If you configure Crash Reporting via AD Policy to capture crash
4186 reports within the domain, then you will receive this output.
4188 * Generate MiniDumps independent of the Windows Crash Reporting.
4190 (a) If an exception occurs,a minidump will be generated at the path
4191 %TEMP%\afsd.dmp. The type of minidump defaults to Normal but
4192 can be set to any other type via use of the
4195 [HKLM\SOFTWARE\OpenAFS\Client] MiniDumpType (DWORD)
4197 (b) MiniDumps can be generated at any time using the "fs minidump"
4198 command. This allows you to generate a minidump without
4199 stopping the process or even requiring the use of a debugger
4202 * an RX connection which reports bad ticket now treats it the same way
4203 the expired ticket is treated. The ticket is removed from the
4204 connection and a new anonymous connection is established.
4206 * fs memdump now requires that the caller be in the AFS Client Admins
4209 * additional debugging information is provided in cm_Analyze when the
4210 error code is ignored.
4212 * Fix a regression introduced into the pioctl() client code which
4213 will cause a crash in winlogon.exe, explorer.exe and fs.exe if
4214 MIT KFW is not installed. (1.3.82a)
4216 * Remove AFS Gateway option from OpenAFS Control Panel (1.3.82a)
4219 * Fix a race condition in cm_EndCallbackGrantingCall() that
4220 could leave 'nrevp' pointing to freed memory. In the process,
4221 optimize the processing to ensure that the applications monitoring
4222 the status of the stat cache entry are only notified at most
4225 * Conditionalize access to the Explorer Shell AFS->Mount Point
4226 ->Remove menu item. Disable the item if the selected item(s)
4227 are not mount points.
4229 * Activate AFS RPC (RX) Free Packet Queues stored in Thread
4230 Local Storage. This eliminates thread contention on the
4231 global RX free packet queue. This should improve performance
4232 on multi-CPU systems.
4234 * Fix the IP address filtering code to properly load/unload
4235 the IPHelper DLL so that Windows XP/2003 and beyond systems
4236 do not have to manually probe the registry.
4238 * CleanupACLEnt() was not being called consistently with
4239 the cm_scache_t object referred to by the back pointer
4240 mutex locked. This could in very rare conditions lead to
4241 an invalid memory access.
4243 * Added a script command to the msi installer "afs_replace"
4244 which can be used to replace one version of OpenAFS with
4245 another without requiring an uninstall and reinstall if
4246 the installed msi is no longer in its original location.
4248 * Apparently the problem with multi-domain forests with cross-
4249 realm trusts to non-Windows realms was not entirely solved.
4250 The authentication to the AFS SMB service failed because
4251 the wrong name was being used. Using ASU as an example,
4252 the authentication was being performed with the name
4253 "QAAD\user" (an account in the forest root) and not
4254 "user@ASU.EDU (the MIT Kerberos principal used to login with)
4256 The solution was to add an additional dependency on KFW
4257 in order or to be able to easily obtain the client principal
4258 name stored in the MSLSA ccache TGT. This information is
4259 used in two locations:
4261 - the pioctl() function
4263 - a new WinLogon Event Handler for the "logon" event.
4265 The pioctl function will now be able to use the correct
4266 name when calling WNetAddConnection2() and the "logon"
4267 event handler will now be able to call WNetAddConnection2().
4268 The hope is that the "logon" event handler will be called
4269 before the profile is loaded but I have not guarrantee
4273 * Fixed a locking error in cm_TryBulkStat() which had the
4274 potential of deadlocking the system for the length of time
4275 it takes to perform a bulk status fetch operation.
4277 * Replaced time conversion code (UnixTime <-> FILETIME) to
4278 be completely arithmetic instead of relying on a bizarre
4279 algorithm involving a variety of C RTL time functions.
4280 This has the side effect that UnixTime and FILETIME which
4281 are both stored in UTC are interpretted as UTC throughout
4282 the year. Windows will apply the same localization to AFS
4283 as it does to NTFS. Applications which rely on the ability
4284 to sync files between the two file systems will no longer
4285 see the timestamps of files in AFS change an hour relative
4286 to the files stored in NTFS or Windows based backup devices.
4288 * Fixed a invalid memory access under a bizarre circumstance.
4289 Windows will allow a physical mass media device to be
4290 installed via Plug N Play to the system and will assign it
4291 the lowest available drive letter. This is true even if the
4292 drive letter is currently assigned in the user session to
4293 a network device via NET USE (or its equivalent.) When this
4294 happens, queries sent to the CIFS server will contain invalid
4295 data. This invalid data was not being caught by the AFS
4296 Client Service and was resulting in a crash when Freelance
4299 * Fixed a reference count error when registering callbacks
4300 on a stat cache object if the callback was already registered
4302 * Add a case to cm_Analyze() in order to handle the case where
4303 cm_GetConnByMServers() returns CM_ERROR_NOSUCHVOLUME because
4304 the server list for the volume is empty. In this case, force
4305 an update of the volume info and retry.
4307 * Insert a missing cm_EndCallbackGrantingCall() which could
4308 result in threads waiting for a callback to complete to never
4311 * In the persistent cache, there is a maximum number of volume
4312 entries. Allow volume entries to be reused if the maximum
4313 number have been allocated and their reference count is zero.
4315 * If we already have a dead virtual connection object, don't
4316 ignore the fact that additional dead vc objects must be taken
4319 * Removed a deadlock condition introduced in the summer of 2004.
4320 Do not hold mx locks around calls to RX functions including
4321 rx_NewCall(). That is what reference counts on the rx_connection
4324 * Fixed an initialization error in afslogon.dll which could
4325 result in random behavior including a failure to terminate
4326 the NPLogon function. (1.3.80b)
4328 * Fixed an error preventing the use of SMB authentication
4329 on Windows 2000. (1.3.80a)
4332 * Updated CellServDB to Public CellServDB 16 Mar 2005
4334 * Fixed DNS AFSDB queries to ensure that "csail" cannot be
4335 misinterpretted as "csail.mit.edu" when the resolver
4336 is configured to append ".mit.edu" to failed DNS queries.
4338 * Fixed another case in which the client would replace connections
4339 to the host when it was not necessary. In this case the
4340 situation would occur if cryptall was on and the connections
4341 were unauthenticated due to lack of an appropriate token.
4343 * OpenAFS for Windows has failed to work at sites which are
4344 utilizing a cross-realm trust between an MIT/Heimdal realm
4345 and a multi-domain Windows forest when the workstation being
4346 accessed is not located in the root domain. This is caused
4347 by a bug in the workstation which was triggered after the
4348 introduction of Windows 2003 Server. When the bug is triggered,
4349 the workstation attempts to authenticate users to afsd_service.exe
4350 by contacting the Domain Controller instead of using the
4351 LSA loopback authentication mechanism.
4353 One of the reasons this bug occurs is because the workstation
4354 does not have a reliable way of knowing that the service whose
4355 netbios name is "AFS" is located on the workstation. This will
4356 be fixed starting in Longhorn Beta 1 by Microsoft. The
4357 "BackConnectionHostNames" registry value will be used to
4358 indicate that the authentications to that service name should
4359 be performed using the loopback authentication mechanism.
4361 In the meantime, when Logon Caching is enabled, we can force
4362 afsd_service.exe to authenticate using the logon cache before
4363 contacting the Domain Controller. This will work with both
4364 password and smart card based logons.
4366 * The allDown logic in cm_ConnByMServers() was wrong. The allDown
4367 flag should not be cleared if a volume's server reference is
4368 marked as "offline". In the case where all of the volume's
4369 servers are either "down" or the volumes are "offline", we want
4370 cm_Analyze() to process the condition as CM_ERROR_ALLOFFLINE
4371 instead of as CM_ERROR_TIMEDOUT. In fact, CM_ERROR_TIMEDOUT
4372 should never occur in practice.
4374 In the case of CM_ERROR_ALLOFFLINE, cm_Analyze() will sleep for
4375 5 seconds, clear the server down and volume busy flags, and
4376 then force an update from the VLDB. This allows the client to
4377 update the location of a volume if the reason for it being
4378 marked offline is because it is being moved. Calls to
4379 cm_ConnByMServers() will be retried either until success or
4380 the RDRtimeout period is reached.
4382 * Correct the Power Management code behavior in response to
4383 Standby, Suspend, and Shutdown events. Instead of flushing the
4384 buffers associated with the mounted SMB submounts, simply write
4385 all buffers which are marked dirty.
4387 * Added support for "TheseCells" to afslogon.dll. "TheseCells"
4388 provides a list of cells other than the default cell for which
4389 tokens should be obtained using the default Kerberos principal.
4390 This functionality is logon domain specific and is only available
4391 when using KFW for authentication.
4393 * Fixed FindNext Invalid Handle error caused by over agressive
4394 attempts at garbage collecting dirSearch entries when the
4395 dirSearch ID wraps from the maximum value to 1.
4397 * Add support for registry defined server preferences for VLDB and
4398 File servers. See registry.txt for details.
4400 * Increased default cache size to 96MB and default number of cache
4403 * Fixed refCount leaks related to directory lookups and pioctl calls.
4405 * Callbacks revoked during a race condition with an attempt to
4406 obtain the same callback no longer result in an inaccessible
4409 * New command line tool:
4411 afsdacl : Set or reset the DACL to allow starting or stopping
4412 the afsd service by any ordinary user.
4414 Usage : afsdacl [-set | -reset] [-show]
4415 -set : Sets the DACL
4416 -reset : Reset the DACL
4417 -show : Show current DACL (SDSF)
4419 * IP addresses are no longer queried once at startup. Instead IP
4420 addresses are obtained as needed. Loopback adapter addresses are
4421 no longer published to the server.
4423 * Pay attention to the MIT KFW registry configuration for automatic
4424 importation of MSLSA credentials. (SOFTWARE\MIT\Leash32,MsLsaImport)
4426 * Fix (once again) case-sensitive comparisons which was apparently
4429 * Activate support for vos listvol -format
4431 * Population of the "AFS Client Admins" group failed on non-English
4432 versions of Windows because the name of the "Administrators" group
4433 is localized. Now we lookup the name of the group by using the
4434 Administrators Group Alias Relative ID.
4436 * Multiple cell token acquisition within afscreds.exe was broken in the
4437 case of Kerberos 5 cross realm authentication. Instead of contacting
4438 the KDC associated with the cell's realm, afscreds would obtain a
4439 token from the KDC of the user's realm. This would result in
4440 "invalid kvno errors" while authenticating to the AFS servers.
4441 Unauthenticated access would work. The symptoms would vary based upon
4442 whether or not the VLDB servers had been contacted using unauthenticated
4443 connections prior to the user obtaining tokens.
4445 * The list of ACL entries was becoming corrupted. It appears as if
4446 an ACL when it expires was not being appropriately cleaned up. In
4447 fact, it was left in the list of ACLs associated with the scache
4450 * Changed the default @sys name list to "x86_win32 i386_w2k i386_nt40"
4451 for 32-bit x86 systems. The default for itanium will be "ia64_win64"
4452 and "amd64_win64" for amd 64-bit processors.
4454 * When executing executables, Windows provides a mechanisms by which
4455 the normal search for DLLs can be bypassed. If a file foo.exe is
4456 being executed and there exists a file or directory "foo.exe.local"
4457 (the contents are ignored if a file), then Windows will search for
4458 DLLs first in the "foo.exe.local" directory and second in the
4459 directory in which the "foo.exe" file is located.
4461 Previous releases of OAFW would improperly return
4462 STATUS_NOT_A_DIRECTORY instead of the expected
4463 STATUS_OBJECT_PATH_NOT_FOUND. This would cause the Windows SMB client
4464 to terminate the search for the DLL causing the execution of the
4465 application to fail.
4467 * Rework the reference counting of smb_vc_t objects. The references
4468 stored in the waiting locks were not counted. This could result in
4469 an assertion if the reference count drops to zero.
4471 * "fs wscell" returns "Freelance.Local.Root" for the workstation cell
4472 when running in freelance mode.
4474 * Added support for persistent caching of file contents, stat entries,
4475 volume data, cell data, and name lookup data. When the registry
4476 value, "NonPersistentCaching", is not defined or set to 0 the file
4477 specified by "CachePath" becomes a persistent cache file. The
4478 size of the cache file is computed at run time. It has a maximum
4479 size of approximately 1.2GB. The contents of the file will be
4480 validated according to the rules specified by the "ValidateCache"
4483 * Added support for UUIDs. UUIDs are kept for as long as the
4484 cache file is intact.
4486 * Added cmdebug.exe and support cache manager debugging callback
4487 interfaces in afsd_service.exe.
4489 Usage: cmdebug -servers <server machine> [-port <IP port>] [-long]
4490 [-addrs] [-cache] [-help]
4491 Where: -long print all info
4492 -addrs print only host interfaces
4493 -cache print only cache configuration
4495 * Symlinks to \\AFS\[all\]... will now be treated the same as
4496 symlinks to /afs/.... However, please use /afs/... as otherwise
4497 the symlinks won't work on Unix.
4499 * Correct a problem with local CellID allocation for cells whose
4500 server list is obtained via DNS instead of a CellServDB file. If
4501 the DNS information expires the CellID assigned to the entry will
4502 be changed. This causes all of the cm_scache_t objects which refer
4503 to the old cellID number to become useless. Attempts to access
4504 files or directories with cm_scache_t objects using the old cellID
4505 will fail since the server list cannot be obtained.
4507 * Correct deadlock condition in cm_EndCallbackGrantingCall which can
4508 be triggered if the volume referenced by the cm_scache_t object is
4511 * The AFS Service needs to respond to SERVICE_ACCEPT_SHUTDOWN messages
4512 in addition to SERVICE_ACCEPT_STOP.
4514 * Move RPC shutdown until after the SMB and RX shutdown procedures
4515 complete. Block until RPC shutdown is complete.
4517 * Modify afslogon.dll (integrated logon) to wait for service if its
4518 state is START_PENDING. If the timeout period occurs, reset to
4519 the retry interval and not the sleep interval.
4521 * When renewing the server list for a cell obtained via DNS AFSDB
4522 records, the cm_cell_t entry must be removed from the list of all
4523 cells. Otherwise, the list of cells will be corrupted.
4525 * In the dcache and scache modules, use the cm_scache_t dataVersion
4526 instead of the cm_fakeDirVersion.
4528 * Synchronize fs.c with the unix version.
4530 * The variable used to determine whether a file or virtual memory
4531 mapped cache is used was not properly initialized to a default
4532 value. If the registry setting "NonPersistentCaching" was not
4533 set, the choice would be random. Properly initialized to be
4536 * The memory mapped view was never unmapped before closing the file
4537 at service shutdown. This is now properly cleaned up.
4539 * Default location of Cache file is now %TEMP%\AFSCache
4543 * A bug affecting new installations of 1.3.75/76 would result in
4544 the creating of incorrect mountpoints in the freelance root.afs
4545 volume for the default cell.
4547 If "fs \\afs\all\<cellname>" lists a volume name of "root" instead
4548 of "root.cell", you have been affected by the bug. To correct the
4549 problem, execute the following commands:
4551 fs rmmount \\afs\all\<cellname>
4552 fs rmmount \\afs\all\.<cellname>
4554 fs mkmount \\afs\all\<cellname> root.cell <cellname>
4555 fs mkmount \\afs\all\.<cellname> root.cell <cellname> -rw
4559 * A bug has been fixed in the auto-generation of Freelance root.afs
4560 symlinks which produced random entries in the \\AFS\all directory.
4562 * Support has been added for multi-homed servers
4565 * Added a new registry value, "StoreAnsiFilenames", which can be used
4566 to force the use of ANSI character sets instead of OEM Code Pages.
4567 This feature is useful when users require the ability to create
4568 filenames with 8-bit characters and need to access the files from
4569 both Latin-1 based Unix systems as well as from Windows.
4571 Activation of this feature will prevent access to files stored with
4572 8-bit OEM characters.
4574 * Shutdown all SMB threads in a synchronized manner when stopping the
4577 * There is currently a maximum cache size of 1.3GB. The limit is imposed
4578 by the largest contiguous block of unused memory within the 2GB process
4579 space which can be assigned to the memory mapped file. Unfortunately,
4580 when the executable digital signature verification code is activated
4581 Windows sees fit to further segment the process memory which in turn
4582 reduces the size of the maximum cache file to less then 800MB. If
4583 larger cache sizes are desired, a new registry value should be set:
4585 HKLM\SOFTWARE\OpenAFS\Client (DWORD) "VerifyServiceSignature" = 0x0
4587 Setting this value will disable the runtime verification of digital
4588 signatures on afsd_service.exe and the afs dlls which it loads. It
4589 will not disable the the version number check on those same files.
4590 The signature verification is not a security messure and is only meant
4591 to enhance the ability to afsd_service.exe to detect potential
4592 destablizing mixtures of DLLs from incompatible distributions.
4594 Added code to auto-disable the signature verification check if
4595 the desired cache size is greater then 700MB.
4597 * Windows' WinTrustVerify(WIN_SPUB_ACTION_PUBLISHED_SOFTWARE) is
4598 used to verify the validity of the afsd_service.exe binary
4599 as well as each of the AFS DLLs loaded by the service. Not only
4600 must the digital signature be valid but the signatures of the
4601 DLL must be signed by the same entity as the service.
4603 * Implement new functions: cm_freelanceMountPointExists and
4604 cm_freelanceSymlinkExists. Use them along with other validity
4605 checks in cm_freelanceAddMount and cm_freelanceAddSymlink to
4606 ensure that name collisions do not occur and that empty strings
4607 are not valid file names.
4609 A symlink may not have a name which would resolve to a valid
4610 cell name. Doing so would prevent access to the cell.
4612 * Add missing cm_HoldSCacheNoLock call to Freelance mount point
4613 re-initialization code. The reference counts of the fake root.afs
4614 volume scache object(s) would become invalid when the mount point
4615 or symlink lists were altered.
4617 * Add registry entries to provide mappings from the afsdsbmt.ini
4618 to the new locations for applications which count on the use
4619 of the old Profile file APIs. These apps are likely to fail
4620 if the user does not have administrator privileges and the
4621 registry is locked down.
4623 * The afs_config.exe submounts dialog had two errors.
4624 First, attempts to remove entries failed because the registry
4625 key was being opened without KEY_WRITE privileges.
4626 Second, when editing a submount entry, changing the name
4627 would add a new key and leave the original one in place.
4628 Now the original submount will be removed if its name is
4631 * In recent months there have been several incidents in which
4632 users have experienced problems starting or accessing
4633 afsd_service.exe and after significant effort has been spent
4634 it has turned out that they have two versions of AFS on the
4635 machine or an inconsistent set of DLLs.
4637 Code has now been added to afsd_service.exe which will walk
4638 the list of modules loaded by afsd_service.exe and validate
4639 that the version of the AFS DLLs matches the version of the
4640 afsd_service.exe executable. If they do not match the service
4643 * When Freelance mode is enabled and there is no registry
4644 key HKLM\SOFTWARE\OpenAFS\Client\Freelance, afsd_service.exe
4645 will attempt to import the afs_freelance.ini file contents.
4646 If the file does not exist, it was creating a dummy file
4647 with a r/o and r/w entry for the default cell and then
4648 importing those values.
4650 This process has been changed. The temporary file is no
4651 longer created. Also, both the OpenAFS Client install
4652 directory as well as %WINDIR% are checked for previous
4653 afs_freelance.ini files.
4655 * Added support for VL_GetEntryByNameN(). Still need to add
4656 support for VL_GetEntryByNameU() for multi-homed support.
4658 * Fix a deadlock situation in afscreds.exe when canceling an
4659 auto-generated Obtain Tokens dialog
4662 * Fix the locking of objects during Directory Searches in the
4663 SMB/CIFS server. The failure to properly lock the reference
4664 counts was resulting in the premature freeing of smb_dirSearch_t
4665 objects while they were still in use by the SMB/CIFS client.
4666 This does not solve the "Invalid Handle" problem.
4668 * Fix Find Cell By Name pioctl call to return a valid cell
4669 name for the Freelance fake root.afs volume.
4670 "Freelance.Local.Root".
4672 * Fix the Explorer Shell Extension Symlinks->Add operation.
4673 The dialog template was missing and the link destination
4674 string was too short.
4676 * Add support for symlinks to Freelance root.afs volume
4677 Stored at HKLM\SOFTWARE\OpenAFS\Client\Freelance\Symlinks
4678 <number> = "<linkname>:<relative-path>."
4679 Use symlink.exe to create, list, or remove
4681 * Remove the fallback to the use of KFW's KRB4 library when obtaining
4682 tokens. We never obtain KRB4 tickets.
4684 * Fix AFS Client Configuration Control Panel to support new SysName
4687 * Fix a bug in afsd_service.exe which could result in the SysName
4688 not being read from the registry.
4691 * Add code to block the issuance of AFS tokens by aklog.exe or
4692 afscreds.exe when the Kerberos 5 principal name contains a dot.
4694 * Modify the IsAdmin() function to always treat the local SYSTEM
4695 account as an AFS client administrator. Affects fs.exe and
4698 * Modify the internal handling of Quota Exceeded errors
4700 * Upgrade all reference count fields in the Windows cache manager
4701 and the osi library to use unsigned long instead of signed short.
4702 A similar fix has been applied to the afs rpc (rx) library.
4704 * fix the Windows cache manager to prevent it from replacing the
4705 rx_connection object associated with the cm_conn_t object on each
4706 and every operation if "fs crypt" was set. This explains the
4707 dramatic performance difference when crypt is used vs clear.
4708 The problem: 'cryptall', a boolean flag indicating whether or not
4709 "fs crypt" is set, was being compared to the rx_connection
4710 cryptlevel which is either rxkad_clear:0 or rxkad_crypt:2.
4711 1 != 2 and therefore the rx_connection was always destroyed
4712 and replaced on each and every operation.
4714 Lock the cm_conn_t object around every call to RXAFS_xxxx functions.
4715 It is not safe for the cm_conn_t object to not be locked because
4716 rx_DestroyConnection might be called from another thread if:
4717 - the user's tokens have changed or expired
4718 - the crypt mode has changed
4720 This fix appears to have also taken care of the problems associated
4721 with Overlapped Writes resulting in Delayed Write errors.
4723 * fix NSIS installer's AdminGroup.exe to properly create and
4724 remove groups when given -create or -remove. The string comparison
4727 * fs sysname now accepts a list of sysname values
4729 * added a new registry value HKLM\SOFTWARE\OpenAFS\Client "IoctlDebug"
4730 DWORD which when set to a non-zero value will cause error message
4731 text to be output to stderr from the pioctl() routine. Useful in
4732 debugging failures of fs.exe, tokens.exe, etc.
4734 * added a test to the power management code to only perform a
4735 flush operation if there is at least one network adapter which
4736 is not a loopback adapter.
4738 * Fix bug in loading of registry value HKLM\SOFTWARE\OpenAFS\Client
4739 "EnableKFW". This value will not be read if the key
4740 HKCU\SOFTWARE\OpenAFS\Client exists; even if the "EnableKFW"
4741 value under that key does not.
4743 * provide mechanisms to force the use of krb524d for Kerberos 5
4744 ticket to AFS token conversion. For afslogon.dll and afscreds.exe
4745 there is a new registry value "Use524" and for aklog.exe a new
4746 command line parameter "-m".
4748 * Fix the pattern matching algorithm to properly match patterns
4751 * smb_ReceiveCoreRename() was factored to produce smb_Rename()
4752 which is used by both the original function and the new
4753 smb_ReceiveNTRename(). smb_ReceiveNTRename() supports the
4754 creation of HardLinks in addition to Renaming. smb_Link()
4755 is a new function which creates HardLinks via cm_Link().
4756 cm_Link() is a new vnodeops function which creates links
4759 smb_ReceiveNTRename() does not support the File Copy and
4760 Move Cluster Information operations described in its interface.
4761 ReceiveNTRename is under documented in CIFS-TR-1p00_FINAL.pdf.
4763 * When opening files via symlinks, we should follow the symlinks
4764 until we reach the actual file stat cache entry. The stat cache
4765 entry of the file should then be stored in the FID instead of
4766 stat scache entry of the symlink.
4768 * return bad operation errors for all unimplemented functions
4769 even if we do not know the functions exist.
4771 * Log bad packets and unknown operation packets to the trace log
4773 * Map CM_ERROR_BADOP to STATUS_NOT_SUPPORTED instead of
4776 * Update list of known CIFS operations to include all those listed
4777 in CIFS-TR-1p00_FINAL.pdf.
4779 * Modify the handling of HKLM\SOFTWARE\OpenAFS\Client\Submounts
4780 to support the REG_EXPAND_SZ type.
4783 * A new Windows authorization group "AFS Client Admins" is now
4784 created and populated with the members of the "Administrators"
4785 group. The group is used to determine which accounts on the
4786 machine may be used to modify the AFS Client Configuration via
4787 the UI and command line tools. afs_config.exe, fs.exe,
4789 * Modify the WinLogon Logoff Event Handler to query NT4 domain
4790 controllers for the remote profile path if Active Directory
4791 services are not available.
4793 * Fix aklog.exe to not add the AFS ID to the username
4795 * PTS registration of new users to foreign cells has been added to
4798 * The cm_Daemon thread is used to perform checks for
4799 down servers, up servers, volumes, callback expirations,
4800 lock maintenance and token expiration. Due to a gaff in
4801 larger integer division the thread never performed any
4802 work. Instead the current time computation would always
4803 be less then the trigger times. This had an adverse affect
4804 on the client's ability to maintain communication with servers,
4805 keep volumes up to date, and flush user tokens and acls
4806 when they have expired. This was broken when the 1.3 branch
4807 was modified to support VC7 which no longer included
4810 * An initialization problem with the Freelance code was
4811 detected while fixing the callbackRequest. The cm_rootSCachep
4812 object is obtained during afsd_InitDaemons() but the callback
4813 information is incomplete. The callback information will not
4814 be obtained until cm_MergeStatus is called from within
4815 cm_GetCallback. Unfortunately, cm_SyncOp did not properly
4816 test for the conditions under which the callback information
4819 * Reports have been filed indicating that callbacks were
4820 being lost. An examination of the code indicated that the
4821 cm_server_t objects were not being properly reference
4822 counted by the cm_scache_t and cm_callbackRequest_t objects.
4823 In particular, the cm_server_t objects may have been freed
4824 from beneath the cm_conn_t objects.
4826 All of the reference counting is now done via the functions:
4831 this improves the ability to track the referrals.
4833 Each cm_BeginCallbackGranting Call now allocates a reference
4834 to the cm_server_t. The cm_EndCallbackGrantingCall either
4835 frees the reference or transfers it to the cm_scache_t
4836 cbServerp field. These are then appropriately tracked
4837 through the cm_Analyze call.
4839 * Ensure that the dnlc hash table is the same size as the
4840 dir name hash table (as per original author's note).
4841 Increase the dnlc CM_AFSNCNAMESIZE to a multiple of 8
4842 for compatibility with 64-bit systems.
4844 * fix smb_ApplyV3DirListPatches to properly apply the hidden
4845 attribute to dotfiles when the infoLevel < 0x101 and
4846 cm_SyncOp has failed.
4848 * Fix the Freelance registry initialization code. There
4849 was a possibility that some systems could end up with
4850 garbage in the registry during a clean install.
4853 * file and directory names beginning with "." will now be given the
4854 hidden attribute when the volume access is anonymous. this matches
4855 the behavior when the volume access is via an authenticated user.
4857 * Added a change monitor to the HKLM\SOFTWARE\OpenAFS\Client\Freelance
4858 key. When a change occurs mark the root.afs data as invalid and
4859 for it to be reloaded on the next access. This allows administrators
4860 to modify the mount point list without restarting the service.
4862 The freelance client used to provide a fake modification time for
4863 the root.afs volume data and its mount points of 7/09/2001 14:24 EDT.
4864 Added code to extract the last modification time of the Freelance
4865 registry key and use that instead. The time now represents the
4866 most recent mount point change.
4868 * PTS registration of new users to foreign cells has been added to
4871 * Additional Cache Control and Credential Manager options have been
4872 added to the WiX installer. See deployment guide for details.
4874 * The CachePath setting is now optionally a REG_EXPAND_SZ type
4876 * The WiX installer has been upgraded. Version 2.0.1927.1 is now
4879 * The loopback installation code may have had a problem updating the
4880 %ETC%\HOSTS file which could have resulted in a premature failure.
4881 Work around code has been added for the case where the file cannot
4884 * The default max chunksize was increased from 15 (32K) to 17 (128K)
4885 because Windows sends 64K blocks when using overlapped writes.
4887 * The default number of server threads was increased from 4 to 25 to
4888 better handle overlapped writes.
4890 * The "AfscredsShortcutParams" registry value was not being properly
4891 loaded by afscreds.exe. Therefore, the default value was always being
4892 used instead of the value set by the installer.
4894 * Windows XP provides downgrade attack detection to prevent an attacker
4895 from being able to force the use of NTLM simply by disrupting
4896 communication with the KDC. This attack cannot exist between the
4897 Windows CIFS client and the AFS Client Service. Therefore, when a
4898 downgrade has been detected the afs pioctl library will force the
4899 establishment of a new CIFS connection using NTLM.
4901 * A locking error was discovered surrounding all references to volume
4902 server lists within the cm_cell.c source file.
4904 * The logged into Windows username was incorrect on Terminal Server
4907 * A new registry value "NonPersistentCaching" was added to the service
4908 parameters key. When set to a non-zero value, the afs cache is stored
4909 in the Windows paging file. There are two limitations to choosing
4911 1. when persistent caching is implemented it won't work with
4912 this flag set since there will be nothing to persist.
4913 2. with this flag set the initial paging allocation cannot be
4914 changed while the service is running
4916 * An initialization bug was discovered in aklog.exe which affected users
4917 who have a domain name for their afs servers which could not be mapped
4921 * afs_config.exe now validates cell names against DNS in addition
4922 to the CellServDB file.
4924 * In order to allow the freelance client to connect to a volume with ID
4925 equal to 1 on the default cell we changed the fake root.afs volume ID
4926 once again. This time we choose 0xFFFFFFFF. In addition, we change
4927 the cell ID of the fake root.afs volume from 1 to 0xFFFFFFFF as well.
4928 It will now be impossible for a volume ID to match that of another
4929 cell unless the client is connected to 0xFFFFFFFD cells. That should
4930 be enough room for growth.
4932 * Fix "fs mkmount" command to work with UNC paths and when
4933 started from non-AFS drives. It is now possible to create a mount
4934 point in the freelance fake root.afs volume with the command
4936 fs mkmount \\AFS\all\<directory-name> <volume-name> <cellname>
4940 fs mkmount \\AFS\all\openafs.org root.cell openafs.org
4941 fs mkmount \\AFS\all\.openafs.org root.cell openafs.org -rw
4943 * The algorithm used to re-attempt access to the servers associated with
4944 a volume has been altered to properly address the case in which all
4945 servers have been marked down. The previous algorithm did not reset
4946 the server's down flags so the servers were never actually retried.
4947 This caused a problem with active volumes if the network connectivity
4948 was lost as could be the case with a network cable removal, wireless
4949 drop, or laptop hibernation. With the fix volume access is restored
4950 almost instantenously when network connectivity becomes available.
4952 * Support for SMB/CIFS browsing has been added to the AFS Client Service
4953 SMB server. It is now possible to use "NET VIEW \\AFS" to obtain a
4954 listing of AFS submounts and freelance mount points. Support for
4955 NETSHAREENUM, NETSHAREGETINFO, NETSERVERENUM2, NETSERVERGETINFO
4956 significantly enhances the behavior of AFS volumes within the Explorer
4957 Shell. For instance, "AFS" now shows up as server in the Explorer
4958 with each submount or freelance mount point visible as a share.
4959 The right click menu in each folder now works with full functionality
4960 on a consistent basis.
4962 * The network provider can be configured to have different behavior
4963 depending on the domain that the user logs into. These settings are
4964 only relevant when using integrated login. A domain refers to an
4965 Active Directory (AD) domain, a trusted Kerberos (non-AD) realm or the
4966 local machine (i.e. local account logins). The domain name that is
4967 used for selecting the domain would be the domain that is passed into
4968 the NPLogonNotify function of the network provider. (see registry.txt
4971 * Added a new registry value [HKCU\SOFTWARE\OpenAFS\Client]
4972 "Authentication Cell" which may be used to specify a default
4973 authentication cell for afscreds.exe which is different from
4974 the default cell for the AFS Client Service daemon.
4976 * Added a Logoff WinLogon Event Notification function to afslogon.dll.
4977 afslogon.dll moved to %WINDIR%\System32\.
4978 New registry entries added to register the dll for Winlogon events.
4980 The logoff event will now force a call to ktc_ForgetAllTokens()
4981 using the context of the user being logged off as long as the
4982 user's profile is not loaded from within AFS. If the profile
4983 was loaded from AFS we can't release the tokens since the Logoff
4984 event is triggered prior to the profile being written back to
4985 the its source location. This is now performed in an XP SP2
4988 * Windows XP SP2 Internet Connection Firewall interoperability
4991 * The %WINDIR%\afsdsbmt.ini contains four sections:
4992 Submounts, Drive Mappings, Active Maps and CSC Policies.
4993 The Submounts and CSC policies are now stored in the registry under
4994 [HKLM\SOFTWARE\OpenAFS\Client\Submounts]
4995 [HKLM\SOFTWARE\OpenAFS\Client\CSCPolicy]
4996 The Drive Mappings and Active Maps are stored in the registry under
4997 [HKCU\SOFTWARE\OpenAFS\Client\Mappings]
4998 [HKCU\SOFTWARE\OpenAFS\Client\Active Maps]
5000 There is no automatic migration of this data as it would be impossible
5001 to consistently migrate data to user profiles which may not be active
5002 when the machine is updated.
5004 * The %WINDIR%\afs_freelance.ini contains lists of mountpoints for the
5005 fake root.afs volume. For the same reasons as for the cellservdb file,
5006 this information should not be in %WINDIR%. This information is now
5007 kept under the registry key
5008 [HKLM\SOFTWARE\OpenAFS\Client\Freelance]
5010 The data from the afs_freelance.ini file will be automatically
5011 migrated to the registry on first execution of afsd_service.exe
5013 * Keeping the CellServDB file in the location %WINDIR%\afsdcell.ini is
5014 troublesome for several reasons. One, it is confusing for those who
5015 expect the file to be named "CellServDB" instead of "afsdcell.ini".
5016 Two, this file is not a Windows Profile formatted file. Three,
5017 applications should not be reading or writing to %WINDIR%. It causes
5018 problems for Windows Terminal Server.
5020 The new location of CellServDB will be the OpenAFS Client install
5021 directory which is by default C:\Program Files\OpenAFS\Client and can
5022 be determined by querying the registry for
5023 [HKLM\SOFTWARE\TransarcCorporation\AFS Client\CurrentVersion]PathName
5025 The existing afsdcell.ini will be migrated by the NSIS installer.
5026 The Wix installer must still be updated to do the same.
5028 * Change NSIS installer to use DNS by default; to remove Integrated Logon
5029 High Security mode; and to add Terminal Services compatibility registry
5030 entries to allow the OpenAFS tools to find the afsdcell.ini and other
5031 configuration files in %WINDIR%.
5033 * Add support for authenticated SMB connections. This will remove
5034 the need for high security mode in most situations. Both NTLM
5035 and Extended Security (GSS SPNEGO) modes are supported. Effectively,
5036 only NTLM can be used even though Kerberos is now supported. The
5037 reason is that it is not possible to construct a service principal
5038 which is unique to each individual machine.
5040 SMB Extended Auth does not work on XP SP2 unless one of two registry
5041 modifications are made:
5043 (1) To disable the check for matching host names on loopback connections
5044 set this key. This does not require a reboot:
5046 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa]
5047 "DisableLoopbackCheck"=dword:00000001
5049 (2) To add the AFS SMB/CIFS service name to an approved list. This
5050 does require a reboot:
5052 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa\MSV1_0]
5053 "BackConnectionHostNames"=multi-sz "AFS" "MACHINE-AFS"
5055 afsd_service.exe will automatically add the current Netbios Name
5056 to the BackConnectionHostNames list and then temporarily disable
5057 the loopback check for one cycle of startup/shutdown of the service.
5058 We assume most folks do not start/stop without a reboot so this
5059 will be adequate in most cases.
5061 * Fix security hole in afslogon.dll which allowed passwords to be
5062 sent in clear text to the KDC in a misformed principal name.
5064 * Fix cm_GetCell() to properly handle expired dns entries
5067 * If Freelance mode is active and the afs_freelance.ini
5068 file does not exist, do not create an empty file.
5069 Instead create a file containing ro and rw mountpoints
5070 to the default cell using the standard conventions.
5072 * Modify the Freelance support to handle the ability
5073 to create rw mount points in the fake root.afs volume.
5075 * Changed the RPC mechanism used for token setting from
5076 named pipes to local. Use of named pipes can be restored
5077 by setting the environment variable AFS_RPC_PROTSEQ to
5080 Named pipes were required when a Windows 9x system was
5081 using a NT system in gateway mode which is incompatible
5082 with our use of local loopback adapters.
5084 * In afscreds.exe, if a username of the form user@REALM is
5085 specified and no password is specified, do not perform a
5086 kinit operation. Only perform the aklog functionality.
5088 * Add a new registry value which allows the number of processors
5089 on which afsd_service.exe executes to be restricted. Valid
5090 values are 1..numOfProcessors
5092 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
5096 * A second MSI based installer option is now available.
5098 * Fixed Kerberos 5 kinit functionality in afscreds.exe to properly
5099 request tickets for user/instance@REALM instead of just user@REALM
5101 * Modify the Power Management Notify routine to wait for the Hard Dead
5102 timeout period instead of a fixed 19 seconds. With the longer timeout
5103 periods Hibernation and Standby could never succeed when network
5104 connectivity is not available.
5106 * The following fs.exe commands are now restricted to Administrator:
5107 - checkservers with a non-zero timer value
5110 - sysname with a new sysname list
5119 setting the default sysname for a machine should be done via the
5120 registry and not via "fs sysname".
5122 * NSIS installer adds options to install Debugging Symbols
5123 and the Microsoft Loopback Adapter; the user is now also
5124 given the ability to select the afscreds.exe startup options.
5126 * Build system modified to generate symbols for FREE (aka RELEASE)
5127 builds as well as CHECKED (aka DEBUG) builds
5129 * Sites which have a volume ID of 0x20000001 assigned to their
5130 root.cell volumes have been experiencing problems with accessing
5131 the root.cell volume of their cell when Freelance mode has been
5132 active. This was because 0x20000001 was assigned to the fake
5133 root.afs volume created by freelance. The fake volume id is
5134 now set to 0x00000001 to prevent conflicts.
5136 * The timeout logic in the AFS Client Service has been wrong
5137 for sometime. It is based on two different assumptions.
5138 First, the SMB client timeout is a fix value as was the case
5139 with OS/2 Lan Manager. This assumption is incorrect. The
5140 SMB timeout in Windows is a dynamic value computed based upon
5141 a fixed minimum timeout to which is added time based upon the
5142 size of the request and the performance characteristics of
5143 the connection. Second, it is the responsibility of the
5144 SMB Server to enforce the timeout requirements of the client.
5145 This is untrue. The SMB Server cannot be expected to know
5146 the requirements of the client. More importantly, if the
5147 SMB server uses the SMB client timeout as a value to restrict
5148 its behavior as an RX client, the performance characteristics
5149 of the local SMB session would be used to prematurely terminate
5150 WAN connections with significantly different performance
5153 The timeout logic has therefore been modified in the following
5155 . the Lan Manager Workstation (SMB) Session Timeout is used only
5156 as a basis for configuring the Connection Dead Timeout
5157 and Hard Dead Timeout values. The Connection Dead Timeout
5158 must be at least 15 seconds longer than the SMB Timeout
5159 and the Hard Dead Timeout must be at least double the
5160 Connection Dead Timeout.
5161 . New registry entries have been added to allow the Connection
5162 Dead Timeout and Hard Dead Timeout values independent of the
5163 Lan Manager Workstation Session Timeout
5164 . The test to enforce the SMB Client Timeout has been removed.
5166 One of the side-effects of removing the enforcement of the SMB
5167 Client Timeout is that regardless of whether or not the SMB client
5168 is available to receive the response (and how would the SMB server
5169 know) the RX protocol response can be used to update the AFS
5170 Client Service state for ready access by future SMB client
5173 This should be the end of the "Server paused or restarting messages"
5175 * Add support for arbitrary UNC paths to the pioctl() support.
5176 This enables the fs commands as well as the AFS Shell Extension
5177 to work correctly when UNC paths are being used.
5179 * Fix afscreds.exe (by updating afskfw.lib) to search for cells via
5180 DNS if the cell configuration cannot be determined via CellServDB
5182 * Add debug info to test whether CM_BUF_WAITING or CM_SCACHE_WAITING
5183 are ever set more than once at a time
5185 * Fix the management of lists of cm_cell_t structures when using
5186 DNS to lookup cell information. The previous code would fail to
5187 reuse the same cellID for a cell if DNS was used more than once
5188 for a given cell name. When the ttl expired, a single cm_cell_t
5189 could be inserted into the cm_allCellsp list more than once
5190 producing a loop. In addition, the vlServerp list belonging to
5191 the cell was not freed resulting in improper refCounting of the
5194 * Add DNS support to cm_IoctlNewCell() which previous only examined
5197 * Add cm_FreeServer() function and call it from cm_FreeServerList()
5198 to properly garbage collect cm_server_t objects
5200 * Add numVCs variable to smb.c to track the number of smb_vc_t
5201 objects created and use it to initialize the vcID field which
5202 previously was set to 0 in all objects resulting in FindByID
5205 * Fixed DNS lookups to work consistently throughout the OpenAFS
5206 product instead of just from within the afsd_service.exe
5208 * Added a runtime check to ensure that AFS Client Service SMB
5209 Server is accepting connections before attempting to mount
5212 * Read IP addresses for volume servers out of the CellServDB
5213 file if gethostbyname() on the hostname fails.
5215 * Fix getcellconfig() to populate both the Hostnames as well
5216 as IP addresses when loading cell data via DNS
5218 * Increase the Connection Dead Time to 50 from 20 seconds
5219 Increase the Hard Dead Time to 120 from 40 seconds
5220 (matches the Unix values)
5222 * Fixed an assertion validating the number of allocated NCBs
5224 * Fixed the build environment to consistently build for
5225 Windows 2000 and above. (APPVER = 5.0)
5227 * Fixed rx_debug to properly validate the receipt in incoming
5228 data with select() and recvfrom(). Do not copy data out of
5229 the socket buffer unless success is indicated.
5232 * afsd_service.exe will now display a message box to the
5233 desktop when it terminates due to an IP Address Change.
5235 * installer no longer deletes AFS Server configuration data
5238 * installer generates a warning dialog if the RPC service
5239 is not properly configured
5241 * installer compressed with lzma instead of bzip2
5243 * afsd_service.exe shutdown crash solved once and for all
5245 * reference counting of smb_vc_t data structures improved
5247 * name space collision of smb_fid_t event objects corrected
5249 * the output of "fs memdump" is now written to
5250 %WINDIR%\TEMP\afsd_alloc.log
5252 * the file TaAfsApp_1033.dll is now properly installed allowing
5253 the User Manager to start
5255 * a new algorithm is used for computing filename pattern matches
5257 * afscreds.exe now accepts user names containing instance
5260 * Fix the Directory Name Lookup Cache to be case-sensitive.
5261 This is crucial in environments in which a Windows client
5262 is accessing a directory with more than one filename that
5263 differs only by case. If the directory contains "FOO"
5264 and "Foo". You want "DEL Foo" to delete the correct one.
5265 We still have a problem in that "DEL foo" will delete a
5266 random filename. This will be addressed in a future release.
5268 * Fix afscreds.exe -M option (renewMaps) to work when High
5269 Security mode is off. Also, remember to disable the ActiveMap
5270 flag in afsdsbmt.ini when a drive mapping is removed.
5272 * Updates to NSIS installer script. AFS Server configuration
5273 data will not be destroyed on un-install or re-install.
5274 Use a better compression algorithm.
5276 * afslogon.dll now uses KFW to obtain tokens when available
5278 * afslogon.dll when given an all uppercase username will
5279 attempt to authenticate with both the uppercase name
5280 and an all lowercase variation
5282 * DST modification removed. The fix appears to make things
5283 worse after a reboot of the machine.
5285 * fs.exe: added "cscpolicy" which is used to
5286 change client side caching policy for AFS shares
5288 Usage: fs cscpolicy [-share <AFS share>] [-manual] [-programs]
5289 [-documents] [-disable] [-help]
5291 * Several uninitialized variables have been initialized
5293 * It is now possible to obtain tokens using cross realm
5294 Kerberos within afscreds.exe:
5296 user: jaltman@ATHENA.MIT.EDU
5298 Will obtain a cross realm ticket for jaltman/DEMENTIA.ORG@ATHENA.MIT.EDU
5299 will will in turn be used to obtain afs@DEMENTIA.ORG.
5300 The resulting token will be stored with the display name
5301 jaltman@ATHENA.MIT.EDU@dementia.org
5303 * aklog.exe has been added to the client
5305 Usage: aklog [-d] [[-cell | -c] cell [-k krb_realm]]
5306 [[-p | -path] pathname]
5310 -d gives debugging information.
5311 krb_realm is the kerberos realm of a cell.
5312 pathname is the name of a directory to which you wish to authenticate.
5313 -noprdb means don't try to determine AFS ID.
5314 -5 or -4 selects whether to use Kerberos V or Kerberos IV.
5315 (default is Kerberos V)
5316 No commandline arguments means authenticate to the local cell.
5319 * All of the resource files have been restructured to adhere to
5320 a set of rules IBM implemented for loading string resources.
5321 These rules had either been forgotten or were not discovered
5322 by folks working on the OpenAFS sources. The end result was
5323 memory corruption. This is primary item which was preventing
5324 the AFS Server from working.
5326 * Increased the size of the maximum ticket size stored in a token
5327 from 344 bytes to 12,000. Increased the buffers used to convey
5328 messages between the pioctl() caller and the SMB Server from
5329 1000 bytes to 12,512. The code appeared to have been writing
5330 above the top of the stack by quite a few number of bytes.
5331 (The increased ticket size is necessary for the next item.)
5333 * When obtaining AFS Tokens via KFW, krb524 is no longer required.
5334 Instead the raw Kerberos 5 ticket is used in its entirety. This
5335 is extremely important as it allows us to use pure Kerberos 5 KDCs
5336 as the source of the AFS authentication. The use of up to 12,000 byte
5337 tickets will allow tickets produced by all versions of Microsoft
5338 Active Directory to be used.
5339 - create a user account.
5340 - designate it DES only
5342 - specify its UPN to be "afs@realm"
5343 - assign a SPN of "afs/cellname" to the UPN with setspn.exe
5345 * Do not enforce the funky 8dot3 pattern matching rule that the first "."
5346 is special when using long file names. (you must use "*.*" and not "*")
5347 Instead only enforce it when performing 8dot3 searches.
5349 * Fixed the DST problem with creation times being set one hour ahead
5351 * Fixed the problem when using \\afs\cell-alias. For example,
5352 \\afs\uncc instead of \\afs\uncc.edu. Do not a new cell struct
5353 for the alias name; instead simply expand the name. One of the
5354 symptoms of this problem was a loss of acquired tokens.
5356 * Fixed the AFS Shell Extension. The Symbolic Link menu was empty
5357 of strings. (Only English strings provided.)
5359 * Fixed the installer to properly replace in use files.
5361 * Fixed the build system to cleanup generated component version files
5363 * The release build compiled with MSVC 6.0 compiler to avoid the
5364 afsd_service.exe shutdown crash. This does not solve the problem
5365 but simply avoids it for the time being.
5369 * fix afslogon.dll to not corrupt memory when High Security mode
5372 * fix afsd_service.exe to not attempt to restore the stack when
5373 an exception occurs. (not safe in multi-threaded programs)
5375 * fix uninstaller to properly remove the CRT and MFC DLLs
5377 * remove a Message Box from afscreds.exe when getcellconfig()
5378 fails on a kerberos realm which is not a cell
5380 The following is a list of changes to the OpenAFS for Window client
5383 * "fs setserverprefs" will leave afsd service deadlocked
5385 * "vos listaddrs" will core dump
5387 * installer sets the appropriate keys to support Integrated Logon
5389 * installer disables the "Find Lana by Name" functionality as it
5390 was causing headaches for many users
5392 * fix the intermittent crash of the power management thread when
5393 shutting down the AFS Client Service
5395 * optimizes the obtain drive mount list functionality which is
5396 executed every time the mount tab in afscreds.exe and afs_config.exe
5397 are refreshed. (this happens a lot)
5399 * fix the service shutdown logic. add the STOP_PENDING state
5400 and do not accept additional service events after we declare
5403 The following is a list of changes to the OpenAFS for Window client
5406 * flexelint was run against the source tree and hundreds (perhaps
5407 thousands) of corrections were applied to ensure prototypes
5408 were in use; types were used consistently; variables were
5409 initialized; unused variables were removed; etc.
5411 * A wide variety of instrumentation was added including the
5412 ability to produce a stack trace from within afsd_service.exe
5415 * Dynamic configuration of the RDRtimeout value based upon the
5416 LanMan Workstation Session Timeout
5418 * The mount root no longer needs to be called "/afs". This
5419 is now set by a registry value "MountRoot" within the key
5420 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5422 * The cell list is now only read out of afsdcell.ini when the
5423 file changes instead of each time a cell is resolved.
5425 * Thread synchronization was added to cm_server.c and ktc_nt.c
5427 * All calls to GlobalAlloc()/GlobalFree() were replaced with
5428 calloc()/free(). The Global functions were needed on Windows 3.x
5429 but have caused a variety of problems on the Win32 platforms.
5430 Avoiding them is highly recommended by several Microsoft
5431 Knowledgebase articles
5433 * Support for Symbolic Links added to the AFS Shell Extension
5435 * Added a registry value "OverlayEnabled" to determine if
5436 Shell Extension Overlays should be enabled.
5437 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5439 * New Build system to support VC6, VC.NET, VC.NET2003 compilers and
5440 separate trees for checked and free builds. Build system supports
5441 a custom directory src\WINNT\extra which can be used as a grafting
5442 location of organization specific additions to the build tree.
5444 * New installer built using NSIS 2.0.
5446 * Named all kernel objects in order to allow them to be monitored
5447 with tools such as SysInternals' ProcExp.exe.
5449 * Introduced new EventLog framework for AFSD
5451 * Introduced Power Management interface to AFSD for Standby and
5452 Hibernate modes to allow cache to be flushed prior to network
5455 * Utilize Win32 DNSQuery API instead of internal routines. This
5456 allows DNS SRV queries to be sent to all current domain name
5457 servers. Not just one specified in an INI file. DNS is now
5460 * "NetbiosName" registry value may be used to specify a fixed
5461 Netbios Name such as "AFS" to be used instead of "HOSTNAME-AFS"
5462 when the loopback adapter is in use. If you need to use the
5463 old notation with a loopback adapter installed specify a registry
5466 "NetbiosName" REG_EXPAND_SZ = "%COMPUTERNAME%-AFS"
5468 * Refactor all modules which depend on LAN Adapter and NetbiosName
5469 determination in a new library: lanahelper.lib. This allows for
5470 consistent behavior throughout the product.
5472 * Move the afsd.log and afsd_init.log files to the directory specified
5473 by the "TEMP" environment variable. This is usually %WINDIR%\TEMP
5474 for services. Added the Date to the log entries.
5476 * New registry value "RxMaxMTU" used to limit the size of the RX
5477 packets sent by the AFS Client Service to the Server. In order
5478 to enable OpenAFS to work across the Cisco IPSec VPN the packet
5479 size must be restricted to 1264 or smaller. The latest NSIS
5480 installer sets a value of 1260 by default.
5482 * New registry value "RxNoJumbo" to disable the use of Jumbo Rx
5483 packets. This is not needed in order to work across the Cisco
5484 VPN but might be needed for other network environments. This
5485 value is not set by the NSIS installer.
5487 * New registry value "HideDotFiles" is used to apply the Hidden
5488 attribute to files whose names begin with a '.'. This value
5489 is set by the NSIS installer.
5491 * New registry value "MaxMpxRequests" allows the maximum number
5492 of multiplexed sessions to be configured at run time. This
5493 value is not set by the NSIS installer. The default value is
5496 * New registry value "MaxVCPerServer" allows the maxmimum number
5497 of VCs per server to be configured at run time. This value is
5498 not set by the NSIS installer. The default value is 100.
5500 * New registry value "AllSubmount" allows the "all" submount to
5501 be disabled by setting its value to 0x00.
5503 * Allow cells names to be valid mount points
5504 \\<netbiosName>\<cellname>
5506 * Store the active state of drive mappings in order for afscreds.exe
5507 to restore them upon startup
5509 * Add exception handling to generate a Stack Trace to the afsd_init.log
5510 file if one happens to occur.
5512 * Add lots of logging to help detect the cause of invalid SMB packets
5514 * Enable Kerberos for Windows to be used to obtain AFS Tokens via
5515 conversion of Kerberos 5 "afs" service tickets. Supports auto-
5516 renewal of expiring tokens as long as afscreds.exe is running.
5518 * New afscreds.exe command line options:
5520 -M = renew drive maps
5521 -N = ip address change detection
5524 * New registry value "EnableKFW" in {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5525 determines whether or not MIT Kerberos for Windows should be used
5526 to obtain tokens via Kerberos 5 tickets.
5528 * New registry value "AfscredsShortcutParams" in
5529 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5530 determines the command line parameters to be specified when "fixing"
5531 the AFS Shortcut in the user's startup folder.
5533 * The "ShowTrayIcon" registry value has been moved from
5534 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5535 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5537 * The <cell name> registry values used to store the token expiration
5538 reminders have been moved from
5539 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5540 {HKCU,HKLM}SOFTWARE\OpenAFS\Client\Reminders
5542 * Obtain the Logon User Name from the Explorer key when available
5544 * new text document doc\txt\winnotes\registry.txt lists all registry
5545 values used by OpenAFS (excluding the AFS Server)
5547 * BUG: rx_securityClass objects were not properly reference
5548 counted and were never freed.
5550 * BUG: reduce the number of conditions under which CM_ERROR_TIMEOUT
5551 would be generated. The existence of a server does not imply
5552 that it is not down. If all of the servers for a cell are down
5553 return CM_ERROR_NOSUCHVOLUME instead. This prevents the Explorer
5556 * BUG: the directory name lookup cache failed to free the entries
5557 in the cache when the name cache entries cycled. The entries
5558 in the cache would become dereferenced without being freed.
5560 * BUG: fs setserverprefs could be executed without Administrator
5563 * BUG: the number of allocated NCB objects (100) exceeded the number
5564 which could actually be waited upon by the kernel (64). Any objects
5565 which were utilized above the limit could never have event completions
5568 * BUG: smb_username_t objects were not being reference counted and
5569 were not properly freed.
5571 * BUG: smb_tid_t objects could under unusual circumstances be freed
5572 before they were no longer referenced.
5574 * BUG: smb_fid_t object pointer were frequently used even when
5575 their value could be NULL. They were not properly released and
5576 therefore they were never freed.
5578 * BUG: smb_packet_t data structures were not completely initialized
5581 * BUG: when Rx produces a CM_ERROR_NOIPC error do not return "Access
5582 Denied" because that causes the Explorer Shell to try again until
5583 access is obtained. Instead return "Remote Resources" which allows
5584 the shell to move on and treat the error as transient.
5586 * BUG: when initializing the NCBreturns structure, separate Event objects
5587 were created for each NCB although a single Event object was supposed
5588 to be shared by all.
5590 * BUG: smb_dirSearch_t objects were not being properly referenced counted
5593 * BUG: smb_tran2Packet_t objects were not being properly referenced
5596 * BUG: directory path creation did not handle the case of multiple
5597 directories requiring creation in one attempt
5599 * BUG: SMB requests which required an Extended Response were ignored.
5600 This prevented some files from being written to AFS volumes.
5602 * BUG: character strings were being freed even after they were
5603 inserted into in use data structures
5605 * BUG: inconsistent usernames were used when High Security mode was
5606 enabled. (there is still much to do in this area)
5608 * BUG: pioctl() calls which require out of band RPC operations were
5609 susceptible to race conditions when performed by multiple processes
5611 * BUG: memory allocation and deallocation crossed instances of the
5612 C Runtime Library producing memory leakage and corruption in
5613 afscreds and the client configurator.