1 Since 1.5.9907 (1.6pre7)
3 * Add Shutdown message to the windows event log
5 * Check Offline Volume status according to the specified policy
6 not every time the daemon thread runs.
8 * if the directory object cannot be found in the tree, return
9 CM_ERROR_PATHNOTFOUND instead of crashing.
11 * avoid memoryt leak of cm_GetSpace() allocations
13 * improve afs/kfw error message output. Include PT and KTC
14 errors. Impacts afslogon.dll, afscreds.exe and afssrvadm.exe
16 * afslogon.dll NPNotifyLogon will now kick start the
17 afsd_service if it is not already in a pending start or
20 * properly construct the userrealm in the afskfw library.
21 do not include the '@' in the realm.
23 * fix the generation of HTML reference manual pages from
26 * No longer send release lock RPCs to a file server for
27 a deleted file. Each attempt will fail with a VNOVNODE
28 abort. Multiple aborts can result in the file server
29 throttling the client.
31 * cm_BkgDaemon should not perform the job of cm_SyncOp
32 which already properly serializes operations on the
35 * cm_BkgDaemon should not perform queued tasks on deleted
36 objects. Doing so is a waste of resources.
38 * Avoid rx nat pings prior to connection attachment.
40 * After directory enumeration completes move the directory
41 status object to the most recently used position to avoid
42 it being recycled when the directory contains more objects
43 that the status cache can hold.
45 * Reduce memory fragmentation caused by lock order validation
48 * When a status object is deleted, move its object to the
49 least recently used position in the queue to promote
50 recycling when the reference count drops to zero.
52 * Fix the osi_Log macros so they safe for use in if..else
53 statements by wrapping in do..while(0). Several logged
54 messages were wrong as a result and there may have been
55 other logic errors caused by mismatched if and else
56 statements when braces were not used in the code.
58 * Apply interlocked operations to all state and queue fields
59 in the cm_call, cm_volume, cm_scache, and cm_buf objects
60 in order to ensure that operations are atomic. State flag
61 bits were being lost on multi-processor systems under heavy
64 Since 1.5.9906 (1.6pre6)
65 * add support for detection of NTFS symlinks to
66 \\AFS using GetFileInformationByHandleEx which is
67 available on Vista, Server 2008 and beyond.
69 * improve thread safety of afs_shl_ext.dll
71 * MS11-043 adds response validation for SMB_COM_NEGOTIATE
72 messages received by the SMB Redirector. OpenAFS failed
73 to properly specify a Challenge and DomainName in the
74 response when the security mode is SMB_AUTH_NONE (or share
75 with password). This patchset corrects smb_ReceiveNegotiate()
76 so that it adheres to the protocol specification.
78 * smb_T2SearchDirSingle() must not fail directory search requests
79 for the _._AFS_IOCTL_._ file. Although this file does not actually
80 exist, it is successfully processed by CreateFile operations.
81 Therefore, an explicit search for it should return a valid answer.
83 * cm_SyncOp/cm_SyncOpDone is used to synchronize the RPC processing
84 to ensure that calls which are in conflict cannot occur at the
85 same time but also to ensure that the ordering of operations
86 is consistent. cm_MergeStatus() was in many cases executed after
87 cm_SyncOpDone() removed the synchronization barrier which in turn
88 permitted status information to be applied out of order. Side
89 effects could have included data loss due to client side file
90 truncation. More commonly two StoreData RPCs would have their
91 status information applied out of order forcing the cache manager
92 to invalidate all of the cached data for the file.
94 * cm_NewServer() can result in a call to cm_UpdateVolumeLocation()
95 if a server probe is performed. In order to avoid recursive
96 calls to cm_UpdateVolumeLocation() do not probe new servers from
97 within cm_UpdateVolumeLocation().
99 * Numerous Rx improvements.
101 * Fix out of order lock acquisition in smb_ReceiveV3LockingX()
102 which occurs if the lock acquisition fails.
104 Since 1.5.9905 (1.6pre5)
105 * aklog supports dotted Kerberos v5 principal names.
107 * afskfw library always attempts afs/cell@USER-REALM
109 * afskfw library must test return code from
110 krb5_cc_start_seq_get() or will trigger a null
111 pointer exception when using Heimdal.
113 * lock protected fields must be 32-bit in order
114 to avoid memory overwrite races.
116 Since 1.5.9904 (1.6pre4)
117 * Fix caching of non-existent volumes. The test to
118 trigger an immediate CM_ERROR_NOSUCHVOLUME in
119 cm_UpdateVolumeLocation() was backwards.
121 * Prevent the background daemon from checking the
122 status of non-existent volumes. cm_CheckOfflineVolumes()
123 should skip volume groups with the CM_VOLUMEFLAG_NOEXIST
126 * The afskfw library should return an error immediately
127 if the krb5_32.dll library cannot be loaded. Affects
128 afslogon.dll and afscreds.exe.
130 * No longer depend on leashw32.dll in afskfw library.
132 * NPLogonNotify must provide the user password in all
133 calls to KFW_AFS_get_cred(). It cannot count on a
134 credential cache being preserved between calls. Permits
135 tokens to be acquired for all cells listed in the
136 TheseCells registry value for a domain.
138 * Improve the trace logging from NPLogonNotify().
140 * Avoid a race when writing the cm_scache_t mountPointString
141 when acquiring mount point or symlink target data via
142 cm_GetData(). The race could result in bogus target
145 * Permit the use of des-cbc-md5 and des-cbc-md4 enctypes
146 as DES keys in asetkey.exe.
148 Since 1.5.9903 (1.6pre3)
149 * Fix automatic addition of Freelance cell mount points
152 * Avoid recursive offline volume checks which could cause
153 a crash due to stack exhaustion.
155 Since 1.5.9902 (1.6pre2)
156 * rx calls issued on a busy call channel are now specially
157 handled and logged. All calls are automatically retried.
159 * The vos support for multi-homed servers was incorrectly
160 implemented. The changes have been reverted.
162 * Avoid wasting cm_buf_t objects and associated data buffers
163 for obtaining mountpoint and symlink target strings. Instead
164 read them directly into the cm_scache_t mountPointTarget
167 * Optimize the evaluation of mountpoint and symlink targets
168 by skipping the FetchStatus and a full round-trip if
169 the cache manager does not have valid status information
172 * Fix the abstraction of cm_FreelanceAddMount() by passing
173 in volume names that do not include the trailing dot.
176 [there was no 1.6pre1 for Windows]
178 * vos commands now manipulate servers by UUID and can
179 recognize multi-homed servers.
181 * afs_config will not longer set the Tray Icon State
182 in the registry if the checkbox is not present in
183 the dialog. [RT 128591]
185 * Heimdal's roken utility library has been added
186 as \Program Files\Common\afsroken.dll
188 * When probing servers to determine if they are up or down
189 no longer issue an RXAFS_GetTime RPC to servers that indicated
190 that they do not understand the RXAFS_GetCapabilities RPC.
191 Since they responded it is known that they are up.
193 * AFS Explorer Shell Extension now works from folder
194 backgrounds. Overlays for mount points and symlinks
195 are present in the dll, but are not registered at present
198 * Do not use RankServerInterval registry value as the value for
199 PerformanceTuningInterval.
201 * CellServDB updated to 13 Dec 2010 release from grand.central.org
203 * Add "fs chmod" command and display current mode as part
204 of "fs examine" output.
206 * When the data version of a mountpoint or symlink changes,
207 the target string in the cm_scache_t object must be cleared.
208 Otherwise, the new target will not be queried.
210 * "fs checkservers" now includes vldb servers in the output
211 and only lists multi-homed servers once. A multi-homed
212 server that has at least one up interface is no longer
213 considered to be down.
215 * When asynchronously storing dirty data buffers to the
216 file server ensure that (a) the cm_scache_t object and
217 the cm_buf_t object are for the same File ID so that
218 locking and signalling work properly; and (b) if the
219 FID no longer exists on the file server, do not panic,
220 just discard the buffer.
222 * When processing VNOVOL, VMOVED and VOFFLINE errors perform
223 server comparisons by UUID or address and not simply by
224 cm_server_t pointer. Otherwise, server failover may not
227 * Do not preserve status information for cm_scache_t objects
228 when the issuing server is multi-homed.
230 * Giving up all callbacks when shutting down or suspending
231 the machine is now significantly faster due to the use
232 of an rx_multi implementation. (This functionality is
233 still off by default and must be activated by a registry
236 * Race conditions were possible when updating the state
237 of the cm_volume_t flags and when moving the volumes
238 within the least recently used list.
240 * Ensure that the lanahelper library does not perform a
241 NCBRESET of each lan adapter when enumerating the
242 current network bindings. Correcting this permits OpenAFS
243 to work on Windows 7 when the network adapter settings
246 * Fix creation of mount points and symlinks as \\AFS\xxxx
250 * Reimplement FetchData and StoreData operations to make
251 use of rx_Writev and rx_Readv which is faster than than
252 multiple calls to rx_Write and rx_Read. Throughput
253 improvements of up to 20% are possible.
255 * Many improvements to the rx rpc stack:
259 - Fast recovery is not triggered by packet resend timeouts
261 - Socket errors are now detected and reported
263 - avoid repetitive memory allocations
265 - avoid calling gettimeofday() for every packet
267 - reduced dependency on the call->lock
269 - When a call receive is completed, ack all of the packets
270 immediately. Do not wait for the first response packet
273 - Do not hold the call->lock during rx_Write* and rx_Read*
274 operations when the call fields in use can only be accessed
275 by the application thread.
277 - Reduce the default window size from 128 back to 32 because
278 the current overhead from walking packet queues in the rx
279 listener thread significantly hampers the ability to
280 process incoming packets from the network.
282 - Only backoff the peer timeout once for a series of naks.
284 - Release builds are lean and mean. All debugging and
285 packet tracking code is disabled.
287 * Version number is logged to the Windows Application Event
288 Log as part of the start pending message.
290 * MSI installers now backup and restore configuration
291 data during upgrades.
293 * Rx pmtu discovery has been disabled by default. Use
294 HKLM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
295 RxPMTUDiscovery "DWORD"
296 value to re-activate.
298 * cm_BkgDaemon event evaluation has been optimized to avoid
299 unnecessary overhead.
301 * If a volume is known to be readonly, then locally fail any
302 change request without sending it to the file server. This
303 avoids triggering the file server's abort threshold.
305 * Do not leak cm_volume_t objects from the volume LRU queue.
307 * Ensure that cm_NameI errors are acted upon promptly.
308 There are many cases in the SMB server where an error from cm_NameI()
309 was either ignored or not acted upon until several other operations
310 are performed that could result in the same error being repeated.
311 This is a mistake which did not have negative side effects until
312 additional checks for callback status were added recently.
314 At present, if a CM_ERROR_ACCESS error is returned and ignored,
315 subsequent attempts to operate on the same cm_scache_t will result
316 in additional queries to the file server that will also end in an
317 abort response. This can trigger the file server to delay responses
320 * Fix fs mount point and symlink create operations which were
321 failing due to improper directory separator termination.
323 * RXAFS_GetVolumeStatus can return VNOVOL, VMOVED, etc. In
324 order to process them and update volume state a fid must
325 be passed to cm_Analyze(). Use the volume root fid.
327 * Negative caching for volume lookups that fail with either
328 VL_NOENT or VL_BADNAME. The expiration time is five minutes.
329 This prevents volume lookup storms.
331 * Provide each volume lookup its own two hour lifetime instead
332 of reseting all volume location information every two hours.
333 This prevents the first smb path evaluation after the volume
334 location data has been discarded from potentially causing an
338 * When the SYSTEM account is in use, the SMB connection is
339 not always established as S-1-5-18. Sometimes it uses an
340 anonymous, S-1-5-7, connection. In both cases the SMB
341 authentication name is the nul string. To better identify
342 when an SMB connection is local system, the smb_username_t
343 is now stored as a SID instead of a user name. When a SID
344 is used, a new flag bit is set SMB_USERNAMEFLAG_SID.
346 In smb_SetToken if AFS_PIOCTL_AFSLOGON is set and the user
347 connection is not SYSTEM, the RPC SID is checked. If RPC
348 SID is SYSTEM, that supercedes the SMB authentication.
350 * Replace STATUS_IO_TIMEOUT with RPC_NT_SERVER_TOO_BUSY for
353 * Replace STATUS_PATH_NOT_FOUND with RPC_NT_SERVER_UNAVAILABLE
354 for CM_ERROR_ALLOFFLINE and CM_ERROR_ALLBUSY.
356 * Optimize behavior of buf_CleanAsyncLocked(). Avoid searching
357 for the cm_scache_t object by FID if it is already known.
359 * If the readonly file attribute is set (stored as a unix mode)
360 then a CreateFile operation should fail if the file is opened
361 for DELETE in combination with any other privilege.
363 * If the directory buffer contents are garbage we can crash
364 the service. Add some simple validation checks to ensure
365 that cm_dirEntry_t objects have the correct flag value and
366 that the name strings are not too long.
368 * If the bulkStat errorCode indicates that a particular object
369 is inaccessible due to a VIO error, we must update the server
370 status appropriately in order to permit failover.
372 * An RX_MSGSIZE error is returned by the new PMTU detection
373 code. It is critical that such an error result in a retry of
374 the operation that failed. Otherwise, the PMTU detection can't
375 work and the server will be marked down.
376 Secondly, it is important that such errors not leak to the
377 application layer. Map them to CM_ERROR_RETRY in all cases.
379 * Translate RX_RESTARTING to the same cache manager error code
382 * When logging server volume instance errors to the windows
383 application event log, be sure to log the cell as well.
384 Translating from server ip address is non-trivial. Make it
385 easier for administrators triaging issues to plug the volume
386 and cell info into vos commands.
388 * Since unix mode bits are represented in octal in most cases
389 make sure we log them that way.
391 * The cm_enforceTrailingDot() function failed if there was a
392 trailing dot followed by white space.
394 * Rx: Do not send RX_MSGSIZE if the PMTU size did not change.
397 * A crash could occur if a single letter share name was
398 dynamically evaluated as matching a cell name. This was
399 due to a failure to treat the comparison of two nul
400 strings as identical.
402 * Docbook validation is performed by xmllint during builds.
404 * Volume package bugs in the file server can result in VOFFLINE
405 being returned to the client instead of VNOVOL or VMOVED. As
406 a result the Unix CM treats VOFFLINE the same as VMOVED and VNOVOL.
407 The Windows client has not. As a result, bugs in the file server
408 can cause the Windows client to lose if the volume has in fact
409 been moved to another server.
411 As part of this change, the volume location list is updated prior
412 to the volume status being applied to the server from which the
415 * If the pages to be flushed are from a readonly or backup volume
416 they can't be dirty. In this case there is no need to stabilize
417 the pages before they are locked and recycled.
418 Stablilization is performed on the cm_scache_t object so do not
419 stabilize and unstablize for each cm_buf_t object.
421 * Over the years the processing of the Freelance callbacks have
422 added functionality that behaves much more like FetchStatus checks
423 to a file server. If the data version of the object has changed,
424 get the new data. Given that is the case, we can remove much of
425 the original refresh logic that is rather race prone. Say goodbye
426 to cm_fakeGettingCallback and cm_fakeDirCallback.
428 * When processing a pioctl path with either smb_ParseIoctlPath or
429 smb_ParseIoctlParent, cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK|GETSTATUS)
430 must be called on the cm_scache_t object to ensure that it is up
431 to date before we permit cm_Lookup or other operations to be performed
432 on it. Add the cm_SyncOp() call to smb_ParseIoctlPath and
433 smb_ParseIoctlParent to ensure it is done for all pioctl operations.
435 * The 32-bit tools installer was displaying client configuration wizard
436 pages. Since, the 32-bit tools share configuration with the AFS
437 service, the 32-bit tools installer should not be attempting to change
438 client configuration. Remove said wizard pages from the 32-bit tools
441 * cm_data.rootSCachep is a global pointer to the cm_scache_t that
442 represents the root.afs volume root directory. Throughout the
443 code this cm_scache_t was being used without ensuring that a
444 callback to the volume is in fact valid.
446 Under most circumstances this would not be a problem. However,
447 it is possible for a request to fail due to the lack of a callback
448 at a critical moment. Add a new function cm_rootScachep() that
449 attempts to ensure that a callback is present (if possible) prior
450 to use the cm_scache_t object.
452 * When the contents of the Freelance root directory changes the fake
453 directory buffers are updated and a fakeDirVersion is incremented.
454 The dataVersion of the cm_scache_t object is supposed to be updated
455 on the next access by performing a fake get callback request.
456 Unfortunately, this did not always occur because of a race. If another
457 Freelance object is updated first, the root directory object would
458 never successfully get a fake callback.
460 This patchset ensures that the generation of the fake directory
461 buffer content and the callback are obtained under the same set
462 of locks thereby removing the race.
464 * Enforce in afsd_service that tokens can be set for alternate
465 userids only if the SetToken pioctl is called from the
466 Local System account.
469 * Revise SMB QuerySecurityInfo Response for MS10-020
471 MS10-020 (http://support.microsoft.com/kb/980232) has caused
472 many problems for implementors of SMB 1.0 servers and applications
473 that call GetFileSecurity() without checking the return code to
474 determine if the call succeeded. The gist of the vulnerability
475 was that the SMB redirector would pass any buffer it received
476 to the application regardless of whether or not it was valid.
477 MS10-020 protects the applications by strictly validating the
478 SMB response data structure and the data in the security descriptor
481 The problem for SMB 1.0 server implementors is that there have
482 been at least three different protocol descriptions for
483 NT_TRANSACT_QUERY_SECURITY_DESC published over the last decade
484 and all of them are incomplete. Therefore, just about no one but
485 Microsoft has an SMB 1.0 server implementation that produces the
486 exact out that they are expecting to validate.
488 The end result is that in an attempt to protect applications from
489 crashing due to invalid input being passed in directly caused
490 dozens of applications to crash by not returning any security
491 descriptor data at all. Even when the applications didn't crash
492 they might not have been able to save their data. Cisco WAAS
493 and NetApp DataOnTap systems were most adversely affected and
494 they have had CIFS protocol licenses for many many years.
496 To fix OpenAFS here is what needed to be done:
498 1. Instead of returning a security descriptor that gives ownership
499 to the NUL SID, give it to the Everyone SID and set the flag
500 that states that everyone has full access.
502 2. Validate the input parameters. In particular, check to ensure
503 that the SMB file descriptor is valid and the file has not
506 3. Enforce the maximum output data and parameter counts.
508 4. Handle buffer overflow and buffertoosmall conditions
509 in the manner that Microsoft expects them to be handled.
510 In particular, note that the parameter data which is returned
511 in the SMB Data Region is not counted in the Data Count.
512 Even if MaxData is 0, we can still return parameters values
513 as long as MaxParm is large enough.
515 * Prevent use of AFSCache file contents if mapped to
518 * The Windows version of "fs newcell" did not accept any parameters
519 and behaved quite differently from the Unix version. Instead of
520 permitting new cell information to be added, the Windows version
521 simply forced the existing cell information to be reacquired.
523 This update adds a new pioctl, VIOCNEWCELL2, to support the
524 implementation of a Unix-style "fs newcell". The functionality
525 added here differs from the Unix version in the following ways:
527 1. "fs newcell" with no arguments is still accepted
528 in order to maintain compatibility with prior Windows
531 2. "fs newcell -cell <cell> -dns" instructs the cache manager
532 to add the new cell but obtain the vldb server info from
535 3. "fs newcell -cell <cell> ... -registry" instructs the cache
536 manager to add the new cell and also save the cell configuration
537 data in the registry for use the next time the service restarts.
539 4. The -vlport and -fsport options are accepted although the
540 -fsport value is currently unsupported by the cache manager.
542 * New registry value "FreelanceImportCellServDB" instructs Freelance
543 to create a mount point for every cell name listed within the
546 * Path MTU discovery for Rx is activated.
548 * Rx socket input buffer is converted to a circular buffer.
550 * Fix usage of cm_FreeServerList(). Do not set the server list
551 pointer to NULL after calling cm_FreeServerList(). Doing so
552 can result in a memory leak.
554 * Only enable Rx NAT pings on a single anonymous connection at a
557 * Fix cm_IoctlSkipQueryOptions() buffer management. Prevents a
558 potential read beyond end of memory buffer.
560 * Reduce requested privileges when reading registry CellServDB
561 to the minimum required.
563 * Add support for RPC Pipe Service NetWkstaGetInfo levels
564 101 and 102 which are called on Windows 7 and 2008-R2.
566 * Prevent integer overflow during quota percent used calculation
567 in Explorer Shell Extension (RT 126846)
569 * Generate a meaningful error if "fs listacls" or "fs setacls"
570 are executed on the Freelance root.afs volume.
572 * RXAFS_InlineBulkStat errors must be processed via cm_Analyze.
573 RXAFS_InlineBulkStatus does not return errors such as EACCES,
574 VNOVOL, VNOVNODE, VOFFLINE, VBUSY, VIO, VMOVED, etc. as an RPC return
575 code. Instead they are returned in the status info errorCode field
578 Traditionally, the error associated with the first FID in the query
579 list has been returned to the caller of cm_TryBulkStatRPC().
580 However, the error has never been processed through cm_Analyze()
581 which means that the per-vnode processing for VNOVNODE and the volume
582 global processing for VMOVED, VNOVOL, etc. has never been performed.
583 As a result, failover to other .readonly volume instances cannot occur,
584 volume moves will not be handled, and files that have been deleted
587 This patchset makes the following changes:
589 1. If an inline bulk operation has been performed and the inline
590 errorCode is a volume global error, then that error replaces
591 the RPC return code within the cm_Analyze() processing for
592 the RPC. This will affect whether or not a retry operation
595 2. The variable 'inlinebulk' is reset to 0 at the top of the
596 cm_Analyze() loop in case failover from an inlinebulk capable
597 file to an inlinebulk incapable file server takes place.
599 3. The FID that is passed into cm_Analyze() is not a real fid.
600 Instead it consists of the cell and volume but vnode = 0.
601 This ensures that the error (if any) is not applied to the
604 4. If an inline bulk operation was performed, prior to performing
605 the cm_MergeStatus() operation a vnode a check is made to
606 determine if an error was returned for that vnode. If so,
607 cm_Analyze() is called with no connection, a fake cm_req_t,
608 the fid, and the error. This permits cm_Analyze() processing
609 to be performed on the file.
611 * Show configuration pages for all types of MSI installations
613 The OpenAFS MSI installer wizard used to not show any configuration
614 pages for "Typical" and "Complete" installations. Setting the
615 workstation cell and logon options during installation required
616 selecting the "Custom" option. Many users choose the "Typical" option
617 during installation, and thus would never see the configuration pages.
618 Therefore, for these users, the workstation cell was being set to the
621 This patch makes the workstation cell and logon option configuration
622 pages visible to all types of installations (except silent
623 installations which show no UI).
625 * cm_LookupInternal creates Freelance mount points and symlinks
626 when queries cannot be found in the Freelance root.afs directory.
627 If the search name is a full cell name for which vldb information
628 can be obtained, then a mount point is added. If the search name
629 is a left-most substring or the full cell name with a dot appended
630 to it, then a symlink was created. This approach created a very
631 poluted Freelance name space.
633 This patchset makes the following changes:
635 1. Do not create symlinks with a dot appended to the cellname
637 2. Do not create symlinks where the left-most substring is not
638 a full dot separated component of the cellname.
640 3. Permit lookups to succeed when we would have created a
641 symlink in the past without creating the symlink.
643 * BPlus tree lookups are much faster than searching through
644 the native directory format on Windows because the case sensitive
645 hash tables cannot be used successfully. Permit BPlus trees
646 to be used except when called with cm_BPlusDirFoo as the action
647 function because cm_BPlusDirFoo is used to build the BPlus trees
648 from the native directory format.
650 * Symlinks are ending up in the Freelance root.afs directory that
651 end with a dot. Make sure it cannot happen.
653 * cm_FreelanceAddMount and cm_FreelanceAddSymlink is supposed to
654 return the allocated FID of the entry that was added. However,
655 cm_NameI is called to perform the lookup without forcing an update
656 of the Freelance fake directory. As a result the entry may not be
659 Force an update prior to calling cm_NameI() by using
660 cm_clearLocalMountPointChange() and cm_reInitLocalMountPoints()
663 * The Freelance fake root directory buffers were not zero-filled.
664 This results in random behavior that can cause the service to
665 terminate unexpectedly.
667 * The validation check for the response from the GetVolumeStatus
668 pioctl is incorrect. The response is not simply a VolumeStatus
669 structure but also several C strings appended to it.
671 * When flushing a file, we need to commit the file length changes
672 as well as the dirty buffers. Call cm_FSync instead of buf_CleanVnode
673 which is called by cm_FSync.
675 * Prevent rx_rpc_stats global lock from being a bottleneck in the
680 * Avoid a race when updating cell vldb server lists
681 that can result in a crash.
683 * Avoid a deadlock when managing CM_SCACHESYNC_STOREDATA
684 state operations for directory objects.
686 * Add new Windows Application Event log messages for
687 VBUSY, VRESTARTING, ALL_BUSY, ALL_OFFLINE, and ALL_DOWN.
688 Include message throttling to prevent the same message
689 from being logged repeatedly within a five second window.
691 * Reduce lock contention by waiting for cm_buf_t I/O
692 operations to complete before permitting cm_SetupStoreBIOD
693 to analyze a buffer for inclusion in a BIOD.
695 * Split the cm_buf_t flags field to separate the flags
696 that are protected by the cm_buf_t mutex from those
697 protected by the buf_globalLock. This eliminates the need
698 to hold both locks everytime the flags field is accessed.
699 Both locks were not held in the past resulting in race
700 conditions that could result in deadlocks.
702 * Add "vos setaddrs" command.
704 * Rx library lock contention avoidance between rx_NewCall and
707 * Rx library races due to inconsistent use of rx_connection
708 conn_data_lock to protect the flags field.
710 * Rx library inconsistent use of RX_CALL_TQ_WAIT which could
713 * Rx library must signal transmit queue waiters when flushing.
714 Otherwise, deadlocks can occur.
716 * In cm_UpdateVolumeLocation, avoid searching for a ".readonly"
717 volume on a numeric volume name.
719 * File buffer allocations whose offsets are beyond server EOF
720 should be locally allocated and zero filled. The file server
721 should not be issued a FetchData rpc which is guaranteed to
724 * Enable integrated logon to work with Windows 7/2008 when
725 user logons are performed with a non-Domain Kerberos principal.
727 * Add Protection Error messages to aklog output.
731 * Prevent the Explorer Shell extension from crashing if
732 symlink creation failed. (126406)
734 * A Rx level NAT ping has been implemented.
735 Add NatPingInterval registry value to
736 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
737 to permit Nat Ping to be enabled. The default value is 0 seconds.
739 * When a re-initialization is taking place, be sure to reset
740 cm_noLocalMountPoints to 0 in case someone deletes the "Freelance"
741 registry key out from underneath the service.
743 * Add krb5 error message translation to aklog, afscreds,
744 afslogon.dll, the network identity manager afs provider
747 * Mode bits aren't directly exposed by the Win32 API. We were leaving
748 them to default to 0777 when creating new files and directories.
749 This version introduces two configuration parameters;
750 'UnixModeFileDefault' and 'UnixModeDirDefault' which are DWORD
751 registry entries that are used to set the initial mode bits.
752 If the values are set to 0, then the behavior is identical to what we
755 * Minidump files are now produced with a timestamp appended
758 * An SMB request debugging monitor has been added. When activated
759 the monitor will automatically turn on trace logging if any SMB
760 request has required longer than 60 seconds to complete and will
761 then create a minidump every 60 seconds thereafter until the
766 * Restore use of DNS AFSDB and SRV records by kaserver clients.
770 * Avoid a potential Freelance deadlock during initial execution
771 of afsd_service.exe if the old ini file data has to be
774 * Three rx library corrections. (1) Idle data connection
775 processing could timeout if the send window filled and
776 took longer than the idle data timeout period for the
777 transmit window to re-open. (2) The transmit queue
778 could be emptied prematurely. A required check for the
779 queue being in use was forgotten. (3) The function that
780 is supposed to implement a wait for the transmit queue
781 to cease being busy failed to wait.
784 * Restore use of DNS AFSDB and SRV records which were
785 unintentionally disabled in 1.5.69
788 * Add a context menu to the NetIdMgr AFS Provider
791 * Prevent an empty directory Btree from being created and
792 marked as valid if cm_BPlusDirBuildTree fails.
794 * cm_BPlusEnumAlloc should not fail if the enumeration
795 contains zero entries.
797 * In cm_BPlusXXX functions, return ENOMEM if malloc() fails.
799 * Do not leak the cm_scache_t dirlock if cm_BPlusDirBuildTree
802 * fs examine should report owner and group ids as signed values.
804 * Protect buffers in smb_WriteData from simultaneous writes.
806 * Query network interface MTU values from the registry because
807 the IP Helper GetAdapterAddresses() fails to report a manually
810 * Improve performance by not dropping and reacquiring the rx_call
811 lock in rx_WriteProc* and rx_ReadProc*.
813 * When configuring the rx library, the network interface MTU
814 must be applied to both the maximum send size as well as the
815 maximum receive size as is done when rx_SetMaxMTU is called
818 There was no 1.5.67 release
821 * Convert afsauthent.dll to use the xdr_alloc and xdr_free
822 routines from afsrpc.dll in place of malloc and free.
823 This permits the two libraries to be built with different
824 C runtime libraries without crashing.
826 * Set the DOS Readonly attribute on a file/directory
827 whenever the unix mode combined with the mask 0200
828 is true. Previously there was a discrepency between
829 the mask used for testing for readonly behavior and
830 that used for setting the attribute.
832 * Disable AFSVolSync based .readonly "whole-volume callback"
833 support because the all file servers prior to 1.5.67
834 (and perhaps 1.4.12) do not properly assign a value to
835 the AFSVolSync structure in bulk status RPC responses.
837 * Improve the error output from aklog to output the value
838 from krb5 error_message() if the afs_com_err output
839 indicates an unknown value.
841 * Hold a lock on the cm_scache_t object when the smb
842 server is assigning a new clientModTime value.
844 * Permit custom build version numbers to be set
845 without modifying the src/config/NTMakefile.<platform>
848 * Add support for binary code signing with cross-signed
849 certificates and arbitrary versions of signtool.exe.
851 * Convert VBUSY and VRESTARTING to CM_ERROR_ALLBUSY and
852 do not permit them to be exposed to the smb redirector.
854 * Convert STATUS_TIMEOUT responses to STATUS_IO_TIMEOUT
855 to avoid confusion within the smb redirector.
857 * Fix the byte order assigned to port numbers associated
858 with AFSDB record lookups. They must be network byte
859 order not host byte order.
861 * Add dynamic server ranking based on RPC round trip
865 * Official support for Windows 7 and Server 2008 R2
867 * Improved Rx RTT computation following the lessons
870 * Prevent a file server bug (FetchData returning an
871 invalid length instead of zero) from causing an
872 "unexpected network error" when writing to files.
874 * Promote DNS SRV records as superior to DNS AFSDB
875 records. Support arbitrary port numbers for vldb
878 * Add "fs listacl -cmd" support.
880 * Add AFSVolSync based .readonly "whole-volume callback"
881 support. With this functionality, multiple objects from
882 a .readonly volume can have their status validated by
883 issuing a single RXAFS_FetchStatus RPC.
885 * Remove drive mapping functionality and service start/stop
888 * Remove drive mapping functionality from afs_config.exe.
890 * Adjust SMB error return codes to avoid returning
891 STATUS_TIMEOUT which results in the SMB redirector
894 * Network Identity Manager OpenAFS Provider now provides its
895 own "AFS lock" notification icon to report the status of
896 "have tokens, have no tokens, service not started, service
897 started but inaccessible". Hovering over the icon lists the
898 cells for which tokens exist (if any) and the OpenAFS version
899 number. Double-clicking executes the Network Identity Manager
902 * When merging the current status for an AFS object, ensure
903 that the volume status for that object is set to 'vl_online'.
905 * Permit cm_Analyze to request a retry even when the cm_req_t is
906 flagged CM_REQ_NORETRY if the RPC failed due to RX_CALL_DEAD.
907 The retry will force the use of a new connection.
909 * Add digital signatures to resource only dlls as required by
912 * Prevent pioctl calls from retrying indefinitely when a sharing
913 violation error occurs.
915 * Add {HKLM,HKCU}\SOFTWARE\OpenAFS\Client DWORD "ShowMountTab"
916 to restore access to drive mapping functionality in afscreds.exe
919 * Permit Rx minimum peer timeout, maximum receive window size and
920 maximum send window size to be configured via the registry.
922 * The MSI installer now sets the ARPINSTALLLOCATION as required by
925 * DNS lookups set the port numbers in network byte order within
928 * Replace use of the 16-bit compatibility API WinExec with
929 ShellExecuteEx in afs_config.exe and afscreds.exe. WinExec
930 is incompatible with Windows 7 / Server 2008 R2 UAC
934 * Short circuit background volume checks if a shutdown
935 or suspend request is received.
937 * Avoid contacting the file server in order to allocate
938 an empty buffer beyond the current length of the file.
940 * When a request for streams on a directory or mount
941 point object is received, do not offer a default stream.
942 This was affecting the saving of roaming profiles.
945 * Fix Windows 2000 compatibility. Do not call rand_s()
946 even if it is compatible with the C run time library
949 * Fix a data consistency error between the output of
950 NetWkstaGetInfo and NetServerGetInfo RPCs, specify the Lan
951 workstation group name "AFS", and report server name as
952 "AFS" instead of "\\AFS" when the caller asks for "\\AFS".
954 * Enable executables to be run from \\AFS on Windows 7.
955 Return "Name not found" instead of "File not found" when
956 a directory or file name cannot be found.
958 * Prevent cache manager from marking file server "down" when
959 the data returned in response to either RXAFS_FetchData64
960 or RXAFS_StoreData64 is invalid.
962 * Add pioctl data validation to the AFS Explorer Shell extension.
965 * After 1.5.62 afslogon.dll would no longer crash but it
966 also would not load "domain" specific configurations.
969 * An uninitialized variable was present in the symlink
970 recursion detection loop.
972 * Reverse addition of unique per file GUID during
973 Create/Open responses. Inclusion of the GUID prevents
974 Cygwin applications (and others) from accessing //afs.
976 * Treat "filename::$DATA" as a synonym for "filename".
977 This addresses the recent report of roaming profiles
978 failing to save back to the file server.
980 * When given a choice between an error from rx_Write()
981 or RXAFS_EndStoreData(), use the rx_Write() error.
982 This ensures that over quota and access denied errors
985 * If an error occurs during RXAFS_StoreData operations,
986 apply that error to all of the buffers in the BIOD.
987 This ensures that failures will not be retried if they
988 will only fail again.
990 * Do not perform a DNS lookup for a cellname if the
991 cellname prefix is _._AFS_IOCTL_._. This will avoid
992 DNS lookups when the query is for
993 _._AFS_IOCTL_._.F7E5F580200909010061TTTTNT7TT.{10E39A49-4531-4496-A08E-842D4C440D20}.
995 * Fix Freelance root.afs volume object callback processing.
996 Object synchronization and status merging bugs.
998 * Fix Explorer Shell "invalid parameter" bug introduced
1001 * Log RXKAD errors to the trace log.
1003 * Prevent simultaneous pioctl calls from multiple processes
1004 from stepping on each other's data.
1006 * Prevent simultaneous pioctl calls from crashing the
1009 * Add data validation to pioctl output processing in
1010 ktc_GetToken and ktc_ListTokens.
1012 * Add data validation to pioctl output processing in
1013 fs.exe and symlink.exe.
1016 * IMPORTANT: New support for DCE RPC Services: SRVSVC and WKSSVC.
1017 No longer will browsing \\AFS produce truncated share names.
1019 * Registry specified server preferences did not result in
1020 the CM_SERVERFLAG_PREF_SET flag being applied to the
1023 * Avoid unnecessary DNS lookups of share names as cells
1024 when it is known that the name cannot be a cell name.
1025 Any name that does not contain a dot is skipped.
1027 * When processing Dfs Referral requests, do not return an
1028 error if the server is down or busy; if the volume is
1029 offline; the cell vldb cannot be reached; etc.
1030 These paths are still \\AFS paths and so the client
1031 should not be sent to search elsewhere for resolution.
1033 * CRITICAL: Prevent the cm_Daemon thread from terminating when
1034 the machine enters the suspend state. IF the cm_Daemon thread
1035 dies, there is nothing to execute down server checks.
1037 * Prevent use of smb_StartedLock before initialization which
1038 can result in a panic in the lock package.
1040 * Improve error checking in afslogon.dll in order to prevent
1041 NULL pointer dereferences within GetLogonDomainOptions()
1042 if the Lsa operations fail.
1044 * Add support to the build system to automatically update
1045 a Microsoft Symbol Store.
1047 * Do not access the cm_conn_t after executing the cm_Analyze
1048 loops. Doing so can result in an access to memory that has
1051 * CRITICAL: If a StoreData request offset is not aligned on the
1052 buffer module blockSize, (offset % blockSize) bytes of dirty
1053 data will not be written to the file server.
1055 * CRITICAL: If a StoreData64 request is sent to a file server
1056 that does not support large files (>2GB) and the client is
1057 unaware that the server has no large file support, the first
1058 chunksize worth of data will not be written to the file server.
1059 This bug was introduced in 1.5.3. It affects all IBM AFS file
1060 servers and all OpenAFS file servers older than 1.4.0.
1063 * If a file server becomes inaccessible while the cache manager has
1064 dirty buffers to write, the afsd_service buf_IncrSync thread can
1065 attempt to use 100% of the cpu.
1067 * Fix the locking that protects the cm_cell_t lists
1068 (all, name, id, free).
1070 * Remove cm_cell_t objects from the name and id hash tables before
1071 placing them on the free list.
1073 * Fix "fs newcell" which was broken in 1.5.60.
1075 * Do not attempt to synchronize dirty buffers if the associated
1076 volume is known to be unavailable.
1078 * Log to the Windows Event Log dirty buffers that cannot be written
1079 during the shutdown process.
1081 * Modify behavior of a Freelance mountpoint target that does not
1082 specify a cell. Instead of assuming the target volume is in the
1083 Freelance.Local cell, use the workstation "Cell" specified in the
1084 registry. A mountpoint target of "#root.cell." will now mean the
1085 root.cell volume in the workstation cell for the current session.
1086 If the workstation cell changes from "athena.mit.edu" to
1087 "andrew.cmu.edu", the referenced volume will also change without
1088 requiring that the mount point targets be altered.
1090 * Avoid false infinite loop errors when validating the cm_cell_t
1091 free list during startup.
1093 * Fix two more locations where xdr_free() should be used instead of
1094 free() to permit a checked built of afsd_service.exe to work with
1095 release builds of afsrpc.dll.
1097 * Add cm_FindServerByUuid(). Re-implement RXAFS_InitCallBackState3()
1098 to permit the server Uuid to be used to lookup the server object
1099 and from that determine the cell. This permits callbacks that are
1100 received from alternate addresses to be processed with a known server
1101 object. Previously a request from an unknown server would clear all
1102 callbacks from all cells.
1104 * Fix a bug that prevented optimal performance when using a non-zero
1105 value for 'daemonCheckVolCBInterval'. As a reminder, when
1106 "daemonCheckVolCBInterval" is set to a non-zero value, all .readonly
1107 volume callbacks are automatically renewed 90 minutes before their
1110 * Fix automatic ranking of vldb servers whose values are obtained from
1111 the CellServDB file.
1113 * Add logging for RX CALL DEAD errors which are generated if the remote
1114 host is unreachable.
1116 * Add failover for RX CALL TIMEOUT errors when the volume is readonly
1117 or the call is to a vldb server.
1119 * Add registry based cell search functionality to NetIdMgr,
1120 afs_config.exe, and klog.exe.
1122 * afsconf_GetCellInfo() has been modified to perform gethostbyname()
1123 lookups on the host names in the CellServDB instead of using the
1124 specified IP addresses. This provides aklog, pts, vos, etc. the same
1125 CellServDB behavior that the Windows Cache Manager uses.
1127 * Completely a behind the scenes change. The snprintf() routines have
1128 been completely rewritten.
1130 * When updating the stat cache entry callback of a .readonly object
1131 from the volume group object, update the file server reference to
1132 ensure it matches the most update to date callback.
1134 * Add proper support for processing callbacks from multi-homed file
1135 servers. Instead of comparing servers by cm_server_t pointer,
1136 compare them by UUID when the UUID is known.
1138 * During a shutdown short circuit the offline volume check daemon
1141 * Properly track link counts during cm_Link and cm_Unlink operations.
1143 * Return the error code of RXAFS_FetchData / RXAFS_StoreData in
1144 preference to an error code reported by rx_EndCall.
1146 * Add "PerFileAccessCheck" registry value to permit testing against
1147 experimental file servers that include per-file acl support. This
1148 value is intentionally undocumented. It is not to be used by
1149 production environment deployments.
1151 * Log a warning to trace log if a volume group is marked as a DFS
1152 File Set. The Windows OpenAFS client does not include DFS support.
1154 * Fix a bug introduced in 1.5.60 that prevents the afs netidmgr
1155 provider from obtaining tokens when referrals are in play.
1157 * Add "fs chown" and "fs chgrp" commands to permit the owner and
1158 group of objects stored in AFS to be set from Windows.
1160 * Avoid performing background daemon operations when the machine is
1161 going into suspend mode.
1163 * Perform offline volume checks in most recently used order.
1165 * Prevent crash when a data version for a cache object goes backwards.
1167 * fs setquota, fs setcachesize, vos setfields, and vos create now
1168 accept human readable orders of magnitude. (K, M, G)
1170 * fs listquota fixed to permit large quota sizes to be displayed.
1172 * rx packet resend and data packets sent counts were incorrect.
1174 * Multi-thread safe library versions are now being generated and used.
1175 mtafsubik.lib, mtafsutil.lib, mtafsvldb.lib, mtafsvol.lib.
1177 * Microsoft SMB Redirector (mrxsmb.sys) support for ExtendedSessTimeout
1178 values are now available on XP through Windows 7. Add functionality
1179 to autodetect if such support is present on the machine. If so,
1180 configure it if necessary and dynamically adjust the AFS Rx timeout
1183 * Correct another race condition in the Rx library that could result
1184 in an unexpected panic while freeing the Rx call iovq.
1187 * A fix to the pioctl library to support drive substitution
1188 to UNC paths. (SUBST <d:> <\\afs\cell\path>).
1190 * On April 9th Microsoft released a Hot Fix for Windows Server 2003 SP2
1191 that corrects a deadlock in the smb redirector and also adds new
1192 functionality that permits the AFS SMB server to be given a longer
1193 timeout than is normally the case. New functionality has been added
1194 to configure these additional LanmanWorkstation\Parameter values.
1195 (This functionality has been backported to XP SP3 and is scheduled
1196 to be released on June 5th.)
1198 * The BackConnectionHostNames registry value configuration was broken
1199 when dynamic re-establishment of Netbios Name registrations was added.
1202 * Hidden vos commands are revealed.
1204 * If the "DisableLoopbackCheck" registry value is set, do not unset
1205 it during the same service session.
1207 * Reorganize code that prevents multiple Store operations to the same
1210 * Modify IsPathInAfs test in Explorer Shell Extension and fs.exe to
1211 permit broken symlinks to be treated as being in AFS.
1213 * Fix vos commands that output 64-bit integer values.
1215 * Cygwin Import Libraries are provided in the SDK for all OpenAFS DLLs.
1216 Permits building cygwin applications against OpenAFS libraries.
1218 * OpenAFS Release Notes, Administrator Guide and User Guide now installed
1219 as Windows HTML Help (.CHM) files.
1221 * NSIS installer does a much better job of cleaning up files left
1222 over from previous installs.
1224 * Fix RT#124787, a race condition between "fs flush <dir>", "fs flushvolume",
1225 or "fs flushall" and on-going directory operations that can result in
1226 afsd_service.exe crashing.
1228 * Add support for DNS SRV records in place of AFSDB records.Â
1229 _afs3-vlserver._udp.<cellname>. Priority field is used.Â
1230 Weight is currently ignored.
1232 * Add a method of specifying Client CellServDB information within the
1233 registry that can be used to either override the CellServDB file or
1234 force the use of DNS lookups for a given cell.
1236 The registry schema is as follows:
1238 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\
1239 "LinkedCell" REG_SZ "[cellname]"
1240 "Description" REG_SZ "[comment]"
1241 "ForceDNS" DWORD {0,1}
1243 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\[servername]\
1244 "HostName" REG_SZ "[hostname]" Default: [servername]
1245 "IPv4Address" REG_SZ "[address]" Used only if gethostbyname() fails.
1246 "IPv6Address" REG_SZ "[address]" <future>
1247 "Comment" REG_SZ "[comment]"
1248 "Rank" DWORD "0..65535" Default: 0
1249 "Clone" DWORD "{0,1}" <future: server only>
1250 "vlserver" DWORD "7003" <future>
1251 "ptserver" DWORD ... <future>
1253 ForceDNS is implied non-zero if there are no [servername]
1254 keys under the [cellname] key. Otherwise, ForceDNS is zero.
1255 If [servername] keys are specified and none of them evaluate
1256 to a valid server configuration, the return code is success.
1257 This prevents failover to the CellServDB file or DNS.
1259 Only one of "HostName" or "IPv4Address" is required. "HostName"
1260 is optional if [servername] is the host name.
1262 * Extend registry based CellServDB functionality to
1263 afsconf_GetCellInfo() interface used by aklog and many
1264 other command line utilities.
1266 * libafsconf.dll moved from OpenAFS\Client\Program to OpenAFS\Common
1267 as it is now used by both client and server components.
1270 * PriorityClass of afsd_service.exe process raised to
1271 "High" to match the priority of the system services
1272 that are waiting for afsd_service.exe to respond.
1273 Configurable using [TransarcAFSDaemon\Parameters]
1274 DWORD "PriorityClass" value. (use values from Win32
1277 * Change error reporting of CM_ERROR_BADSHARENAME to
1278 use the correct error (NTSTATUS_BAD_NETWORK_PATH).
1279 The error we were sending would cause the smb
1280 redirector to drop the connection.
1282 * Change the error reporting of CM_ERROR_ALLOFFLINE
1283 and CM_ERROR_ALLDOWN to return NTSTATUS_PATH_NOT_FOUND.
1284 Previous we would return Bad Network Path which
1285 would cause the smb redirector to cancel all file
1286 handles on the same "share".
1288 * A debug version of afsd_service.exe can now be used
1289 with a release version of afsrpc.dll (or vice versa).
1290 afsrpc.dll now implements and exports xdr_free().
1292 * Removed all unprotected OutputDebugString() calls as
1293 those calls can block for up to 10 seconds if there
1294 is contention for the globally shared memory or if
1295 the debugger fails to read data in a timely manner.
1297 * Visual Studio 2008, WDK 6.1 and DDK 6.1 are now
1298 supported for builds. The production releases still
1302 * Discovered that XP/2003 supports a variant on recvmsg
1303 called WSARecvMsg. The input parameters are mostly
1304 compatible with the Posix recvmsg. This will save
1305 multiple memcpy() calls per received packet.
1307 * Discovered that Vista/2008 supports a variant on
1308 sendmsg called WSASendMsg. The input parameters
1309 are mostly compatible with the Posix sendmsg. This
1310 will save multiple memcpy() calls per sent packet.
1312 * Exporting rx_ReadProc32, rx_WriteProc32 from afsrpc.dll
1313 and utilizing rx_ReadProc32 from within the cache manager
1314 as part of the FetchData operations. rx_ReadProc32 and
1315 rx_WriteProc32 are optimized versions of rx_ReadProc and
1316 rx_WriteProc for use when reading 32-bit integer values.
1318 * The Network Identity Manager AFS provider now reports
1319 progress events. These provide a better user experience
1320 with Network Identity Manager 2.0.
1322 * [RT 124293] A race condition exists which permits the
1323 scp field of the an smb_fid_t object to become invalid
1324 after a request on the smb_fid_t has begun. Remove the
1325 race by protecting all access to the scp field with the
1326 smb_fid_t mx mutex and obtaining a local reference on
1327 the cm_scache_t object for the length of the request.
1329 * [RT 124276] If the vldb is out of sync with the contents
1330 of the file servers, afsd_service will retry too many
1331 times when a file server reports a volume as not being
1332 present. Now if the list reported by the vldb is the
1333 same as the previously seen list, then the retry is
1336 * [RT 124276] Read-only volume failover was broken in
1337 1.5.53 whenever accessing a volume results in VNOVOL or
1340 * [RT 124276] Prior to 1.3.70 the volume server reference
1341 list was not reference counted and would be prematurely
1342 freed while in use. When reference counting was added
1343 in 1.3.70 a bug was introduced that could result in
1344 service reference list corruption.
1346 * Add Windows Application Event Log warning messages for
1347 "Client SMB MPX value too large" and
1348 "Client SMB Buffer Size too small".
1350 * Renaming of files across directory boundaries would result
1351 in an invalid handle error when attempting to access the
1352 files after the move.
1354 * Fix the handling of Tran2 Set Path Info RPCs. Do not fail
1355 when a smb file descriptor cannot be found. The whole point
1356 of using a Path Info function is because an smb file
1357 descriptor wasn't allocated.
1359 * More edge cases in which dynamic addition of Freelance
1360 root.afs entries would get the wrong FID or where the
1361 root.afs directory would not be refreshed.
1363 * Buffer overflow could occur if the workstation cell name was
1364 longer than 64 characters. Crashes could occur in
1365 afscreds.exe, afslogon.dll, and afsd_service.exe.
1367 * VNOSERVICE and VOFFLINE errors were leaking and were exposed
1370 * Improve consistency of the bad data version value used for
1373 * Log file server uuid values as part of the cm_server object
1374 when available. Dump the cm_server object list in response
1377 * Optimize the performance of resetting access control lists
1378 when tokens are set or removed.
1380 * Fix vos.exe so that it can work with very large volume
1383 * Remove symlink recursion tests and increase max symlink count
1387 * Undo the "UAC manifest fix" applied to afs_config.exe.
1389 * Fixed some additional race conditions in the Rx library
1391 * Ensure that Freelance allocation of vnodes follow
1392 the AFS convention of odd vnodes are directories
1393 and everything else is an even vnode.
1395 * Add Freelance logic to mount point and symlink
1396 evaluation functions.
1398 * Enhance smb_ParseASCIIBlock() so that it can handle
1399 all of the STRING formats defined by the CIFS Technical
1402 * Validate the output of smb_ParseASCIIBlock() in all
1403 callers. Return CM_ERROR_BADSMB if the STRING field
1404 cannot be parsed. CM_ERROR_BADSMB will cause the
1405 contents of the packet to be logged.
1407 * If multiple SMB Raw Write operations were taking
1408 place at the same time, there could be data corruption
1409 because unique event objects were not generated for
1410 each Netbios receive operation.
1414 * Fix a race in rxi_WriteProc which could result
1415 in a freed or in use packet being passed to
1416 rxi_PrepareSendPacket if the call is currently
1417 in a transmit wait state. Passing a freed packet
1418 will result in a panic. Passing an in use packet
1419 might result in a call failure or perhaps data
1422 * Kerberos referrals broke the NIM AFS provider
1423 logic for automated configuration of the workstation
1424 cell with a new identity. Now if a referrals
1425 response is returned, assume the cell belongs to the
1428 * The interface to buf_CleanAsync() and
1429 buf_CleanAsyncLocked() has been changed to permit
1430 errors to be propagated to the caller.
1432 * During an enumeration of a directory, if the btree
1433 has not yet been generated, return an error.
1435 * When mapping CM_ERROR values to NTSTATUS values
1436 do not treat 0 as an error.
1438 * Avoid a cm_scache_t object reference count underflow
1439 when renaming via the SMB interface.
1441 * smb_IsMask() can be called with a non-8.3 input buffer.
1442 Separate smb_IsMask() into smb_Is8Dot3Mask() for 8.3
1443 input buffers and smb_IsMask() for non-8.3 input buffers.
1445 * Add cm_buf_t error state to the "fs memdump" output.
1447 * pioctl drive mapping checks need to take into account
1448 Global Drive Mappings and SUBST assignments.
1450 * Rename operations should not be permitted if the
1451 destination name collides with an existing name that
1452 differs only by case UNLESS the collision is the object
1453 that is being renamed.
1455 * Increase the maximum number of background daemon threads
1456 to 64. Leave the default at 4.
1458 * Set the FILE_UNICODE_ON_DISK flag when reporting the
1459 file system characteristics.
1461 * Report the maximum component name length as 255 even
1462 for Unicode as part of the file system characteristics.
1464 * Add RX debugging information to "fs memdump" output
1467 * Refcount the cm_server_t objects returned by
1470 * Fix the UAC manifest applied to afs_config.exe
1472 * rxdebug now indicates if any RX_CONN flag is set
1473 not just some of them
1475 * Change the mapping of CM_ERROR_WOULDBLOCK from
1476 NTSTATUS_LOCK_NOT_GRANTED to NTSTATUS_CANT_WAIT
1480 * Includes #GCO Public CellServDB 07 Nov 2008
1482 * Prevent separate cell entries from being created for
1483 "foobar.org" and "foobar.org." which can happen if
1484 an AFSDB record is found instead of a CellServDB
1487 * Fix a panic when checking the SMB virtual circuit list.
1488 If a VC is destroyed while the list is being checked
1489 it is possible for the next object to become invalid.
1491 * Increased logging to afsd_init.log of netbios name
1494 * Fix a DNLC panic when a UTF-16 string that cannot be
1495 converted to a utf-8 string is provided as a filename.
1497 * In the NSIS installer, delete old afslogon.dll/afscpcc.exe
1498 from %windir%\system32 and fix the WinLogon\Notify\AfsLogon
1499 location to refer directly to the new location in
1500 %programfiles%\openafs\client\program
1502 * Be more aggressive about responding to Netbios
1503 NRC_BRIDGE/NRC_NOWILD errors. Rebind the Netbios name to
1504 the LANA within the listener thread without performing a
1507 * Use the WNet Enumeration of all connected drives as a filter
1508 in the pioctl GetIoctlHandle operation. This should remove
1509 the last situations in which there was a delay when opening
1510 an Explorer Shell properties menu with the AFS Shell
1513 * Fix the parsing of CellServDB in Windows to handle linked
1514 cells as is done on Unix.
1516 * Add support for linked cells to the cache manager,
1517 netidmgr afs provider, and aklog. When two cells are
1518 linked, a volume lookup that fails in one cell is looked
1519 up in the other. This can be used for construction of
1520 test cells that override specific volumes; cell renaming;
1523 * Increase the length of cell and realm names that can
1524 be entered in the NIM AFS provider.
1526 * Actively prevent a 32-bit afsd_service.exe from
1527 executing in the WOW64 environment.
1529 * Protect against invalid IP address values in CellServDB
1534 * Fix an Rx panic caused by an attempt to double free
1535 an rx_packet when an error occurs while writing the
1538 * Fix an exception cause by a race condition when
1539 processing multiple file change notifications registered
1540 by the Explorer Shell or other applications.
1542 * Modify the Rx debug interface to use var_args and
1543 permit Rx debugging to be used on 64-bit Windows
1544 which previously had it disabled.
1548 * Automated lock hierarchy enforcement added to the
1549 osi base lock package. This permits detection of lock
1550 violations that might but often does not result in
1551 deadlocks. Enabled by default in the checked builds.
1552 May be activated in release builds via the registry
1553 value "LockOrderValidation" in the TransarcAFSDaemon
1556 When a violation is detection an assertion is thrown
1557 generating a dump file and a message indicating which
1558 locks were obtained out of order is logged to the
1559 Windows Application Event Log.
1561 * Attempt to ensure that a dump file will be generated
1562 whenever the service panics.
1564 * Removed unused lock cm_bufGetMutex
1566 * Many out of order lock acquisitions were detected
1567 using the lock order verifier and were corrected.
1569 * A race condition within the function path cm_GetSCache() ->
1570 cm_GetNewSCache() -> cm_RecycleSCache() permitted a
1571 cm_scache_t object to be issued simultaneously to two
1572 threads. This would eventually result in a panic due
1573 to the resulting undercount.
1575 * Fix interpretation of the empty string as the ioctl path
1576 to mean the current directory. "fs lsm", "symlink list",
1577 etc. now return a "not a ..." error instead of "does not
1580 * Add SMB_STRF_SRCNULTERM flag to smb_ParseStringXXX functions.
1581 to indicate that the input string is a nul terminated string.
1582 Assign ti when input strings are nul terminated instead of
1583 length counted. (from 1.5.50)
1585 * The CIFS protocol specification for handling NT_TRANSACT_CREATE
1586 does not match the observed behavor. The 'nameLength' is specified
1587 in bytes, not in characters. Fix the implementation to match
1588 the observed behavior. (1.5.50)
1590 * The cm_HaveAccessRights() attempt at deadlock avoidance by
1591 calling lock_TryRead() on the parent directory cm_scache_t rw-lock
1592 does not avoid the deadlock. Avoid the deadlock by enforcing the
1593 lock order of the lowest vnode first. Then remove the infinite
1594 loop avoidance in cm_SyncOp() that was returning an unwarranted
1595 access denied error.
1597 * File Server lock synchronization was not propertly enforced in
1598 smb_ReceiveNTCreateX() and smb_ReceiveNTTranCreate().
1600 * Another change to the short name generator. Plus and equal are
1601 not valid characters in a short name according to the
1602 RtlIsNameLegalDOS8Dot3() in kernel mode. They have been replaced
1603 with left and right brace in order to minimize the impact.
1605 * Fix a race condition that can result in a heap overwrite error if
1606 the number of known file servers or vldb servers increases during
1607 the preparation phase of a cm_CheckServers() operation.
1609 * If the SMB Authentication would fail, a LSA allocated memory
1610 buffer would be leaked.
1612 * Rewrite the cm_cell_t allocation code to permit cell vldb server
1613 lookups to fail without leaking cm_cell_t objects.
1615 * Add "fs setcrypt auth" to permit the use of rxkad authentication
1616 only mode. Not documented.
1618 * Initialize Rx mutex and lock objects once and only once as part of
1619 the DLL load process.
1621 * When reporting the Volume Name "AFS" do not include the terminating
1622 NUL in the reported length.
1624 * Refresh Volume Location information upon receipt of a VNOVOL error
1625 from the file server and avoid a race condition that prevented the
1626 updated volume location information from being destroyed immediately
1627 after it was acquired.
1629 * In response to VNOVOL and VMOVED, remove the volume id from the server's
1630 volume list so that the volume's up and down status is not affected
1631 when the machine is shutdown.
1633 * Many fixes to the Rx stack related to packet allocation and deallocation;
1634 Rx debugging; Rx packet tracking across queues; and Rx packet loss caused
1635 by overwriting the rx_call currentPacket pointer when it was already
1636 referring to a packet.
1638 * Rx Jumbograms are now off by default. To re-enable, add "RxJumbo" = "1"
1639 in the TransarcAFSDeamon/Parameters key.
1641 * "fs flush", "fs flushvolume", and "fs flushall" now destroy locally
1642 constructed directory B+ trees for flushed directory objects.
1644 * When cached pages for directory objects have been locally modified, do
1645 not permit the locally modified pages to be mixed with pages subsequently
1646 obtained from the file server. Doing so can result in directory
1649 * If Ubix errors are received in response to VL_xxx operations record
1650 the error on the server instance and permit the operation request to
1651 be retried in case another server is able to answer the query. We have
1652 seen at least one instance of a cell in which a VL server returned
1653 UNOQUORUM (possible database truncation) when the other servers were
1654 responding successfully.
1657 * Optimize the assignment of write errors to dirty
1658 buffers by applying the first error to all buffers
1659 belonging to the same object instead of trying each
1662 * Add resource version info to xstat_fs_test and xstat_cm_test
1664 * Fix a delete vs delete[] usage problem in the WiX custom
1667 * When creating a mount point, validate the volume name to
1668 make sure it exists.
1670 * When processing a mount point, pay attention to VL_BADNAME
1673 * Remove a race condition from cm_volume_t recycling that could
1674 have resulted in two threads making use of the same object at
1677 * Fix a reference count leak on smb_vc_t objects introduced in
1678 1.5.50. Permit the objects to be freed once again.
1680 * Fix the output of "fs memdump" for smb_vc_t objects that was
1683 * Add routines for debugging smb refcount errors.
1685 * When shutting down the cache manager, do not permit AFS
1686 callback requests to be processed. Doing so can result in
1687 invalid memory accesses.
1689 * Do not recycle a deleted cm_scache_t object unless the
1690 refCount is 0. Doing so can result in a deadlock and
1691 invalid memory access.
1693 * Minimize the number of RPCs sent to the file server for
1694 directory bulk status requests. Only request status for
1695 objects that we are actually reporting on and only if we
1696 do not already have status for the object and then make
1697 sure that we return as much as we can at once via the SMB
1698 interface but not so much that it exceeds the RDR timeout
1700 * Enable bos restricted operations
1702 * Fix the creation of submounts used by the afscreds and afs_config
1705 * Avoid repeated retries when attempting to obtain access rights.
1706 If the access rights were obtained successfully but do not
1707 satisfy the request, fail the request.
1709 * Fix error code reported when attempting delete a file on a
1710 readonly volume or one that is marked with the readonly DOS
1713 * Fix a short name truncation bug.
1715 * Fix a heap corruption error when reading the CellServDB file
1718 * Fix directory listings when the directory has "lookup" permission
1719 only and the directory has never been previously accessed.
1721 * Fix several tracking issues for cm_scache_t rw locks.
1723 * Ensure that all buffers used to store cell names are the
1726 * Restore Windows 2000 support (broken in 1.5.50).
1727 LOCALE_INVARIANT was introduced in XP.
1729 * Add RxUdpBufSize registry value. Default is 256K.
1731 * Avoid multiple deadlocks between cm_freelanceLock and cm_scacheLock
1733 * Do not include trailing NULs when returning file names in
1734 a directory search response.
1736 * Prevent RX from crashing the afsd_service by attempting
1737 to allocate to many buffers at once when memory is fragmented.
1739 * Avoid a deadlock when freeing cm_server_t objects
1741 * Optimize search cell file. Do not search to end of file
1742 if the cell has already been found.
1744 * Remove explorer shell extension delay on local disks.
1746 * Automatically pre-allocate 64 Rx packets per thread
1748 * Add smb lock requests and stat cache lock allocations to
1749 the output from fs memdump.
1752 * Fix Large File Support (broken in 1.5.50)
1754 * Fix "fs setacl" (broken in 1.5.50)
1756 * Fix pioctl path evaluation of drive letter
1757 mapped paths. (broken in 1.5.50)
1759 * Fix installation of VC8 Runtime library in the EXE
1760 Retail installer. (broken in 1.5.50)
1762 * Avoid a potential deadlock during Volume Location
1763 updates if all the file servers are unreachable at
1764 the time the VL update succeeds.
1766 * Ensure that rx connection object is references
1769 * Fix propagation of over quota errors during cm_FSync()
1772 Since 1.5.39 [1.5.50 released 16 July 2008]
1773 * There were no releases between 1.5.39 and 1.5.50.
1775 * FIRST RELEASE WITH UNICODE SUPPORT
1777 * Pioctl interfaces to the cache manager refactored
1778 to provide layering between the SMB specific code
1779 and the general purpose ioctl operation.
1781 * Garbage collect dead SMB virtual circuits as soon
1782 as they are no longer being referenced. This avoids
1783 problems with outstanding locks not being dropped
1784 when the virtual circuit becomes invalid.
1786 * Remove the IBM Administration Reference documentation
1787 and replace it with the OpenAFS Command Reference
1790 * Avoid calling rx_SetDeadTime and rx_SetHardDeadTime
1791 functions each time a connection is about to be used.
1792 Do not hold a lock on the rx connection object while
1793 it is being selected. This avoids a race between
1794 threads attempting to set the timeout values and
1795 removes a bottleneck that was hampering performance.
1797 * Ensure that the smb directory attribute is set
1798 for all directory objects.
1800 * Add vs2008 support to the NSIS installer scripts
1802 * Replace the VC Runtime EXE installer with the MSI
1803 installer in the NSIS installer scripts
1805 * Properly delete the VC7.1 C Runtime libraries.
1807 Since 1.5.36 [1.5.39 released 23 June 2008]
1808 * There were no 1.5.37 or 1.5.38 releases for Windows
1810 * Use Visual Studio version of vsprintf() instead of
1811 home grown version that resulted in linking conflicts
1813 * Prevent buserver, ptserver and fileserver from crashing
1814 due to non-portable strftime() format strings.
1816 * Prevent server crashes caused by rx_InitMorePackets()
1817 being called prior to allocation of all of the required
1818 synchronization objects.
1820 * Fixed a memory leak in the internal B+ tree directory.
1821 For objects with non-8.3 names, the 'longname' memory
1822 allocation would be leaked during object deletion or
1825 * Fixed a memory leak in the background daemon cm_CheckServer()
1826 routine. The memory allocated to store the server list
1829 * Obtain a missing lock around a call to cm_RemoveSCacheFromHashTable().
1831 * Correct an abstraction layer violation. cm_scache_t objects
1832 should be marked deleted in cm_Unlink() and cm_RemoveDir() and not
1833 in smb_CloseFID(). Cleanup of deleted cm_scache_t objects should be
1834 performed in cm_ReleaseSCache() when the reference count hits zero.
1836 * Prototype cm_AdjustScacheLRU() and re-implement it using osi_QAddH().
1838 * Do not reference 'smb_logp' in cm_vnodeops.c. Use 'afsd_logp'
1841 * Ensure that scache object 'nextp' pointers are set to NULL
1842 when the objects are removed from the hash table linked list.
1844 * If a dirty buffer cannot be written to the file server because
1845 of an access denied error, mark the error in the cm_buf_t
1846 structure so that it can be dealt with instead of retrying
1850 * Update CellServDB file to GCO Public 23 Apr 2008
1852 * Fix a cm_buf_t reference count leak when attempts to write
1853 dirty buffers to the file server from within cm_IncrSyncer()
1856 * Prevent udebug from crashing.
1858 * Activate RX Idle Timeouts. If the file server is busy for more
1859 then 30 seconds, attempt to failover to another server without
1860 marking the busy server down.
1862 * Another VNOVNODE issue fixed. When writing a dirty buffer
1863 to the file server, if VNOVNODE is received, mark all buffers
1864 as invalid without further attempts to contact the file server.
1866 * Improved performance on high latency links.
1869 * Fix a bug preventing the re-initialization of the Freelance
1870 root.cell contents when a change is made.
1872 * Properly return CM_ERROR_NOSUCHFILE when removing a symlink
1873 or mountpoint that does not exist.
1875 * aklog, afscreds, integrated logon, and the NIM AFS credential
1876 provider will all attempt to first acquire an afs service ticket
1877 using the service principal: afs/<cell>@<CLIENT_REALM>. This
1878 permits the proper detection of the cell realm when multiple
1879 realms are local realms for the cell.
1881 * Fix fs commands that re-use file identifiers instead of forcing
1882 a path evaluation each time. This was broken in 1.5.33 when
1883 hashes were added to the cm_fid_t structure.
1885 * Fix symlink and mountpoint removal and then recreation when
1886 applied to the fake Freelance root.afs volume.
1888 * Fix SMB change notifications for the Freelance root.afs volume
1889 when creating or removing symlinks and mount points.
1891 * Force DNS AFSDB cell name lookups to lower case if the DNS
1892 response is not all lower case.
1894 * 32-bit installers are now built with Visual Studio 2005
1896 * All vos.exe commands now support the "-noresolve" option which
1897 forces the display of IP addresses instead of DNS hostnames.
1899 * No longer hold the cm_volume_t mutex across VL_xxx RPCs. This
1900 permits operations that do not require up to date volume
1901 location data to continue while waiting for the RPC to complete.
1902 Multiple threads waiting for a volume location update on the
1903 same volume set no longer trigger multiple calls VL_xxx RPCs.
1905 * Volume server references could be marked offline and never
1906 reset to the not_busy state.
1908 * Further optimizations to cm_Analyze() retry logic for
1909 CM_ERROR_OFFLINE and CM_ERROR_ALLBUSY error handling.
1912 * Optimize the DNLC by applying Interlocked incrementing for
1913 statistics gathering and enabling greater use of read
1914 locks instead of write locks
1916 * Further optimize osi_Log operations when logging is disabled
1917 by removing the need for the function call overhead.
1919 * Add a new lock operation, Convert Read to Write, and apply
1920 it throughout the cache manager. This function provides a
1921 fast transition from a read lock to a write lock when the
1922 caller is the only reader.
1924 * Fix a timeout problem when opening files or request locks
1925 when the file in question is already write-locked by another
1928 * Fix a deadlock that can occur if "fs flushall" is executed
1929 while applications are writing to AFS.
1931 * The cache manager would re-read data from the file server
1932 after file truncation when obtaining buffers to write to the
1933 same file until the first StoreData RPC completes.
1935 * Fixes error handling in cm_NTCheckDelete()
1937 * Replaces cm_scache_t mutex with a read-write lock permitting
1938 additional parallel access.
1940 * Prevent an cm_scache_t reference undercount when evaluating
1941 symlinks containing @sys when none of the active sysnames
1942 match any of the available targets.
1944 * Prevent the leak of a cm_scache_t rwlock if associated symlink
1947 * Set RxMaxMTU registry value to 0 which is the equivalent of
1948 it not being set at all.
1950 * Reduce contention between BkgDaemon threads
1952 * Prevent CM_SCACHEFLAG_ASYNCSTORE from being reset on a write
1955 * Convert reference counts on cm_volume_t, cm_conn_t, cm_server_t
1956 objects to use the faster Interlocked operations which avoid
1957 the need to hold write locks.
1959 * Permit rxkad errors other than RXKADEXPIRED to be treated as a
1960 non-fatal error that forces a retry against another service
1963 * Fix a crash in smb_WriteData that would occur if an application
1964 opened the file as read-only and then sent a write request anyway.
1967 * Respond to VL_NOENT errors by removing cm_volume_t object and
1968 setting it to be the first object to be recycled. This is not
1969 only a more efficient use of resources but it also prevents
1970 repeated attempts to query the VLDB server for the volume from
1971 the CheckOfflineVolumes routine.
1973 * Change the defaults for RxEnablePeerStats and RxEnableProcessStats
1976 * A first cut at a cache manager statistics monitor that can be used
1977 to determine the necessary cache parameters to support the
1978 application working set.
1980 Off by default, the performance package can be activated by setting
1981 "daemonPerformanceTuningInterval" in the service Parameters key.
1982 As with the other daemon interval values the unit is in seconds.
1984 At service start and each succeeding interval the cache manager
1985 will write statistics to %TEMP%\afsd_performance.txt showing the
1986 relative usage of cm_scache_t, cm_volume_t and cm_buf_t objects.
1987 The FID statistics keep track of all FIDs seen by the cache manager
1988 during the service session whether or not they are backed by any
1989 live object in the cache.
1991 These statistics are not stored in the cache file.
1993 * Fix windows event logging broken in 1.5.33.
1995 * Eliminate multiple calls to syscfg_GetIFInfo() which is used to
1996 populate the IP address list for WhoAreYou/TellMeAboutYourself RPCs
1997 as well as when computing the server preference list.
1999 * When merging status after a StoreData operation prevent data
2000 buffers from being discarded that we would prefer to keep.
2001 (Bug introduced in 1.5.33.)
2003 * Modify installers the set RxMaxMTU to 0.
2005 * In the AFS Configuration Panel on the Advanced->Miscellaneous
2006 dialog, the Background Daemons and Service Threads fields were
2009 * Increase default number of Background Daemons to 4.
2011 * Remove cm_volume_t reference from the cm_scache_t object. Permit
2012 cm_volume_t objects to be recycled.
2014 * Windows Error Reporting has shown a number of previously unseen
2015 stack corruption errors in the Explorer Shell extension. The
2016 minidump appeared to indicate the problem was in ParseAcl().
2017 Reviewing the function it appears that numerous things could go
2018 wrong if invalid input was provided. Added error checking and
2019 data validation. Hopefully this will address the problem.
2021 * Windows Error Reporting has shown crashes in aklog as a result
2022 of krb5_cc_get_default() failures not being handled. They are
2026 * The Rx library used a 32-bit type for sockets which was
2027 truncating the socket value on 64-bit Windows. This
2028 was not an issue in general but became an issue when
2029 Rx "hot thread" support was enabled.
2031 * Enable "Rx hot thread" support which ensures that while
2032 a received message is being processed that a new thread
2033 is allocated to receive the next message to arrive.
2035 * Increased parallel access to smb_rctLock protected
2036 resources. Reduce the number of times that the
2037 lock is dropped and re-obtained.
2039 * Increased parallel access to cm_scache_t and cm_buf_t
2040 resources by replacing the previous global write lock
2041 protected reference counting mechanism with one built
2042 upon the use of InterlockedIncrement and InterlockedDecrement.
2044 * Implement a hash for cm_fid_t to reduce comparison costs
2045 and inline the cm_FidCmp() function.
2047 * Redefine the BUF_HASH and BUF_FILEHASH in terms of the cm_fid_t
2048 hash which has a better distribution
2050 * Modify cm_ConsiderPrefetch to evaluate the amount of data in
2051 the most recent read request instead of the next chunkSize.
2052 cm_chunkSize can be dozens or hundreds of buffers. As a result
2053 too much time is spent performing the evaluation.
2055 * Fix the usage of cm_scache_t bufCreateLock. The purpose of this
2056 lock is to prevent the creation of new buffers while a truncation
2057 is being performed. All references to bufCreateLock have been
2058 removed except in two places:
2059 i. a write-lock surrounding the function that calls buf_Truncate()
2060 ii. a read-lock within buf_GetNewLocked() that actually allocates
2063 * Modify the CM_CONFIG_DATA_MAGIC value to include a
2064 CM_CONFIG_DATA_VERSION number which value be used to force the
2065 replacement of the cache file contents when incompatible changes
2066 are made between releases.
2068 * CM_SCACHESYNC_ASYNCSTORE should not be ordered by
2069 cm_SyncOpCheckContinue
2071 * Avoid calls to multi_Rx if nconns == 0
2073 * Modify smb_WriteData to perform background writes based upon the
2074 crossing of 'smb_AsyncStoreSize' boundaries instead of cm_chunkSize
2075 boundaries. This will slow down writes from the SMB interface but
2076 will avoid the risk of the CIFS client disconnecting from the AFS
2077 client SMB server. The default value for 'smb_AsyncStoreSize'
2080 * SMB AsyncStore functionality is on by default but can be
2081 disabled using the registry value:
2083 HKLM\SOFTWARE\OpenAFS\Client DWORD "EnableSMBAsyncStore"
2085 smb_AsyncStoreSize can be adjusted with
2087 HKLM\SOFTWARE\OpenAFS\Client DWORD "SMBAsyncStoreSize"
2089 The value must be a multiple of the buffer block size
2090 and cannot be larger than the chunk size.
2092 * Do not leak a read lock on the bufCreateLock
2094 * Do not forget to call cm_SyncOpDone if cm_SyncOp succeeded
2097 * In MergeStatus, only remove clean buffers from the hash table.
2099 * Avoid the need to update the data version number on each buffer
2100 associated with a scache when MergeStatus is called after a
2101 StoreData by maintaining a range of valid data versions as part
2102 of the cm_scache_t object.
2104 * Profiling shows large numbers of blocked calls in cm_HoldSCache
2105 from cm_HaveAccess... We can safely avoid the call so do so.
2107 * Add "ForceAnonVLDB" service parameters registry value. Default
2108 is 0. Set to 1 to force the use of anonymous connections to
2109 the VLDB servers. This is useful when using krb5 tokens and the
2110 VLDB servers have not been updated.
2113 * Do not attempt to search for a cell name that matches one
2114 of the special share names "svrsvc", "wkssvc", "ipc$"
2116 * Issue permission denied warnings from afscreds if attempts
2117 to modify the startup shortcut or registry values fail.
2118 This happens under Vista UAC.
2120 * Do not return BAD_NETWORK_ERROR in response to attempts
2121 to create directories or files in the Freelance root.afs
2122 volume. Instead return ACCESS_DENIED.
2124 BAD_NETWORK_ERROR was being sent because the Freelance cell
2125 has no vldb servers. Short circuit the error by testing for
2126 freelance in CreateFile and MakeDir
2128 * Treat srvsvc, wkssvc and ipc$ as pseudo files. Permit them
2129 to be opened and closed without actually existing. For now
2130 we treat them as equivalent to ioctls but this should
2133 * Fix RAP processing to not reject names that exist in the
2134 root.afs volume for GetInfo queries.
2136 * In afscreds, use GetUserNameEx to obtain the username instead
2137 of searching the registry. Try UserPrincipalName first and
2138 fallback to SamCompatible.
2140 * Stop linking to rpcns4.dll which is no longer supported on
2141 Vista and 2008. TaAfsAdmSvr is going to have to be
2144 * When processing dfs referral requests, if the sharename is
2145 a partial match and therefore will not be added automatically
2146 to the share list, make sure that we return an error.
2148 * Avoid a crash when logging VNOVOL errors returned from
2149 cm_CheckOfflineVolume.
2151 * When trace logging is disabled, do not permit osi_LogSaveString
2152 to perform the string copying. By disabling the string
2153 copies, throughput of the service more than doubles.
2155 * when checking offline volumes, update the cell vlserver info
2156 if required and don't perform a check if there are no servers
2159 * Add volumes in the vl_unknown state to the list of objects to be
2160 checked as offline or retried when an allbusy error is received
2163 * Add support for configuring rx_ExtraPackets
2165 * All calls to buf_Get() must be protected by a read lock on the
2166 cm_scache_t bufCreateLock
2168 * When a MergeStatus results in the invalidation of buffers, do
2169 not remove buffers from the hash tables that have active references.
2171 * Upon receiving a VNOVNODE error from a file server, be sure to
2172 invalidate any file locks. This prevents on-going attempts to
2173 extend the locks and will cause subsequent attempts to access
2174 the file to receive a invalid handle error.
2176 * Prevent smb_ReceiveTran2SearchDir from adding terminating NULs
2177 to directory entry names.
2179 * No longer discard tokens when RXKADBADTICKET error is received.
2180 Instead return STATUS_NO_KERB_KEY error.
2183 * Add "cmdebug <host> -cellservdb" option and client service
2184 implementation permits querying the vldb server for each
2185 cell known to the cache manager. The output is in a form
2186 that permits it to be used as a CellServDB file.
2188 * Add Symlinks->Show option to the AFS Explorer Shell Context
2191 * Prevent buffer overrun in the cmd parser used by command
2192 line tools such as fs, vos, pts, etc.
2194 * The path ioctl operations have several issues:
2196 (1) the specified path for "fs examine, whereis, whichcell,
2197 flush" and so always has the follow symlinks and mount
2198 points semantics. This makes it impossible to determine
2199 what the FID of a symlink or mount point is.
2201 (2) "fs examine" out is not the result of a single pioctl
2202 operation but is actually the combined output of half
2203 a dozen operations. Path evaluation is an expensive
2204 operation. It would be faster if the caller could
2205 evaluate the FID first and then perform all of the
2206 rest of the operations by specifying the FID instead
2209 (3) fs output reports all objects as files. By adding a
2210 GetFileType pioctl more informative output can be provided
2211 that indicates what type of object the path evaluates to.
2213 (4) the Windows fs command includes a number of commands that
2214 do nothing but exist only because the Unix cache manager
2217 A new extendible data structure cm_ioctl_query_opts_t has been
2218 added which can be optionally specified with pioctls that do not
2219 already require input data. The first two fields of this structure
2220 are 'literal' and 'fid'. The literal field is used to indicate
2221 whether the last component of the path should be evaluated following
2222 symlinks and mount points.
2224 The fid field permits a fid to be specified.
2226 * A new GetFileType pioctl has been added. The type of objects are
2229 * A new "-literal" option is available for "fs examine, flush, whereis,
2232 * Unimplemented fs commands have been removed.
2234 * There are circumstances where a volume object is being accessed and
2235 the volume is marked indicating that the volume location information
2236 is out of date but where it is also pointless and perhaps dangerous
2237 to block waiting for the rpc to complete. One example is when
2238 processing the cmdebug requests. If we know that we are not going
2239 to use the volume object to contact a server, then we can now set
2240 the CM_GETVOL_FLAG_NO_RESET flag.
2242 * when moving up in the directory tree we search the recorded fid list
2243 to find a matching fid that we have already crossed. we must also
2244 reset the fid count based upon what we discover.
2246 * Cell names are published as share names. As a result they are searched
2247 for as part of the Dfs Referral evaluation. If share "foo" can not be
2248 reached, the CIFS client will fallback to searching for "fo" as well.
2250 Since the freelance client automatically adds a symlink for prefixes
2251 of the cell name, this results in both "foo" and "fo" being added to
2252 the freelance root.afs volume. This patch prevents that negative side
2255 * when tracking Kerberos credential cache names be sure to include
2256 the cache type prefix.
2258 * Microsoft has assigned OpenAFS a network type value
2260 * The algorithms that were used to produce a Netbios name were broken.
2261 The name that was produced when the hostname was too long was in fact
2262 longer than the maximum permitted netbios name.
2264 Also, the "NetbiosName" value was not used as a suffix when the loopback
2265 adapter was not installed. The hardcoded string "AFS" was used instead.
2267 * avoid another deadlock during server probes initiated by the ipaddr
2268 change daemon thread.
2270 * prevent the afs client service from crashing when fs newcell is
2273 * In the NSIS installer, move afslogon.dll, afscpcc.exe and afs_cpa.cpl
2274 to \Program Files\OpenAFS\Client\Program in order to get them out
2275 of the %windir% tree
2277 * EDQUOT == WSAEDQUOT. Define it in the right places so that in the
2278 end STATUS_OUT_OF_QUOTA can be returned to the application.
2280 * Modify the search order for determining the location of CellServDB
2281 and other client configuration files.
2283 1. AFSCONF environment variable
2285 3. NEW - use All Users\AppData\OpenAFS\Client only if CellServDB exists
2286 4. use Program Files\OpenAFS\Client
2288 * More Vista Power Management tweaks
2290 * Close a socket handle leak in the kauth routines
2292 * Don't hold a cm_volume_t mutex across RPCs
2294 * Convert downTimes to use clock time instead of relative times.
2295 This permits correct comparisons with expiration times.
2297 * Update the Wix installer to properly install the prerequisites
2298 for translate_et.exe
2300 * Do not translate pioctl paths as they are not converted by the
2303 * Implement multi_rx probes for checking whether servers are up
2304 or down. Turns probes into a constant time activity instead
2305 of one dependent upon the number of known servers.
2307 * Treat VNOVOL errors the same as VOFFLINE. Do not force an
2308 update of the volume location data and retry. VNOVOL means
2309 that the volume could not be attached. It doesn't mean that
2310 the volume is not on this server.
2312 * Modify the Explorer Shell AFS Context menu Symlink->Add to
2313 not attempt to validate the target path. There is no reliable
2316 * Improve performance of \\afs\<cell><mount-type><volume>\
2317 access by eliminating extraneous attempts to search the
2318 CellServDB file and DNS AFSDB records. Share names
2319 containing '%' or '#' are not valid cell names.
2321 * Return STATUS_RANGE_NOT_LOCKED if the application attempts
2322 to unlock a range that is not currently locked.
2324 * Enable the 32-bit tools to read the 64-bit registry keys
2325 when executing in the WOW64 environment.
2329 * Fix buffer refcount leak introduced in 1.5.29
2331 * Prevent the NIM AFS Provider from crashing when configured to use
2332 Kerberos v4 for token acquisition in situations where no Kerberos v4
2333 support is available. For example, 64-bit KFW or sites that delete
2334 the Kerberos v4 DLL.
2336 * In afsd_init.log, support logging cache sizes greater than 2GB.
2338 * The volume status output of "fs examine" was being determined based
2339 upon the wrong error value. Instead of using the pioctl() return
2340 code, the errno value must be used.
2342 * Prevent invalid pts auto-registration attempts by aklog and the
2343 NIM afs provider when Kerberos referrals are in use. (krb5 1.6+)
2345 * Add synchronization protection to all of the SMB Listener State
2346 variables. This prevents race conditions when the addition or
2347 loss of a network adapter takes place.
2349 * On Vista, there is a race condition between the restart of the
2350 afs client service and the network adapters. Ensure that the
2351 afs client service can handle switching between loopback and
2352 non-loopback modes of operation.
2354 * Fix FollowBackupPath to work with mount points that are not
2355 explicitly read/write.
2357 * Fix a deadlock that has been observed on Vista while resuming
2358 after at least four hours of sleep.
2360 * Add new HKLM registry configuration options that can be used
2361 to provide configuration hints to the NIM AFS Provider when
2362 creating new identities. See release notes for details on
2363 HKLM\SOFTWARE\OpenAFS\Client\Realms\.
2366 * Speed up write buffer operations by holding a global read-lock
2367 instead of a write-lock when updating the buffer data version
2368 number. The data version is protect by the associated stat
2371 * Speed up read buffer operations by removing out of date buffers
2372 from the buffer hash tables.
2374 * Protect against a null pointer dereference in afslogon.dll
2375 if QueryAdHomePathFromSid() fails.
2377 * Remove a race condition while accessing private freelance data
2380 * Update the buffer data versions when Freelance directory data
2383 * Complete the implementation of smb_ReceiveTran2GetDFSReferral().
2386 * Return EAS_NOT_SUPPORTED in response to EA read/store/query
2387 instead of returning zero length.
2389 * Return ERROR_RETRY when a server sends VBUSY.
2391 * Return PATH_NOT_COVERED instead of BADSHARENAME for DfsLinks when
2392 a volstat plug-in is in use.
2394 * Enhance the volstat interface to support Dfs path mappings.
2396 * Fix "symlink list" to do the right thing for Dfs links
2398 * Fix callback expiration that was broken in 1.5.27.
2400 * Fix a crash at shutdown caused by the lack of synchronization
2401 within the daemon threads (cm_Daemon, cm_BkgDaemon, cm_IPAddrDaemon)
2403 * Hold the cm_server_t mutex while walking or modifying the
2404 cm_server_vols_t list in order to avoid a race condition.
2406 * Ensure that the cm_server_t refCount is safely modified.
2408 * The pthread library implementations of pthread_cond_wait
2409 and pthread_cond_timedwait were failing to return with
2410 the mutex held when error conditions occurred. This could
2411 result in race conditions wherever they are used. One case
2412 in which problems were seen was the Rx event_handler().
2414 * Prevent cm_GetBuffer() from locking all buffers associated with
2415 a stat cache object before deciding if the current buffer is up
2418 * Fix cm_BkgPrefetch to actually ensure that there is a cm_buf_t
2419 object for each offset that we will attempt to reserve.
2421 * Reduce the default number of background daemon threads used to
2422 prefetch or store to one in order to prevent two or more threads
2423 from stepping on each other.
2425 * Reduce the default chunk size to 256KB in order to prevent read
2426 timeouts on low bandwidth connections.
2428 * Fix a race condition that could result in threads becoming stranded
2429 when synchronizing multiple requests on the same stat cache object.
2432 * The GiveUpAllCallBacks functionality that was added in 1.5.21
2433 has been disabled. It has been disabled because it turns out
2434 that the RPC is improperly implemented in all file servers
2435 previous to 1.4.6. The RPC is executed without the proper
2436 locks being held resulting in data corruption and eventual file
2437 server failure. The functionality will be re-enabled in a
2440 * a couple of .readonly Volume Callback optimizations were
2442 - Apply most recent volume callback to all stat cache objects
2443 in the volume. This avoids premature callback expirations
2444 and unnecessary FetchStatus calls.
2445 - Add option to permit attempts to renew .readonly Volume
2446 Callbacks on a periodic basis to avoid invalidation of stat
2447 cache data. [Requires daemonCheckVolCBInterval be set to
2448 non-zero number of seconds. 1800 (30 minutes) is suggested.]
2450 * Adds support for 64-bit data version values. Previous releases
2451 only supported 32-bit values. Larger values sent by the file server
2452 would result in rollover.
2454 * aklog and asetkey no longer generate an exception if Kerberos for
2455 Windows is not installed.
2457 * If the LSA authentication functions report a temporary out of memory
2458 condition, fail the current authentication and do not cause the AFS
2459 client service to terminate unexpectedly.
2461 * When the AFS client service is shutting down, prevent the daemon check
2462 thread from executing additional checks which might attempt to access
2463 discarded locks or unmapped memory.
2465 * Correct a double free condition in the NIM AFS credentials provider.
2467 * Add "FollowBackupPath" service parameter registry value.
2468 This provides equivalent functionality to the UNIX afsd -backuptree
2469 option. When set to a non-zero value, normal mount points originating
2470 in a .backup volume will prefer a .backup volume.
2473 * Fix NSIS uninstall from the .MSI installer.
2475 * Add Stack Frame logging for AMD64
2477 * Add support for EWOULDBLOCK to cm_Analyze. If the file
2478 server returns EWOULDBLOCK, retry the request every two
2479 seconds for up to the RDRtimeout.
2481 * Reorganize the order of the includes to ensure that EWOULDBLOCK
2482 is not assigned the same value as EIO.
2484 * Update CellServDB to GCO Public 25 Oct 2007
2486 * Add additional validation and error handling code after
2487 each call to getSlot(). If an invalid slot is returned,
2488 return NONODE. If the invalid slot is returned when
2489 extracting a data node, invalidate the tree.
2491 * Modify compareKeys() to always perform a case-insensitive
2492 comparison and only perform a case sensistive comparison
2493 if the case-insensitive one matches. This ensures the
2494 ordering is consistently reported.
2496 * Add lock assertions to ensure that all calls are being
2497 performed with the correct locks being held. There have been
2498 some crash reports that provide stack data that does not appear
2499 to be possible unless there is a race. However, there are no
2500 obvious locations where the race is taking place and the test
2501 suite indicates that all of the correct locks are being held.
2502 We shall see what happens in the field.
2504 * For consistency replace all calls to findKey in which the range i
2505 s (1,numentries) with calls to getSlot().
2507 * Optimize the depth search loop by testing the slot value in the
2508 for statement instead of forcing the loop to be broken later.
2510 * Reorganize the locking for cm_BeginDirOp and cm_EndDirOp.
2511 There are a number of locations where locks are obtained, dropped,
2512 and reobtained. This reorganization attempts to accomplish several
2515 (1) be optimistic for the most common case so it will be fast
2517 (2) add consistency checks after each location where locks are
2518 dropped and re-obtained. If we lose a race in cm_BeginDirOp
2519 and the bplus tree is out of date, retry until we get to a
2520 consistent state that we can use.
2522 (3) Ensure that all operations take place with the correct locks.
2524 * One of the issues that has become a serious problem since the
2525 addition of the local directory updates is that although cm_SyncOp
2526 synchronizes operations, it does not preserve the order of requests.
2527 This has always been a problem in that it has been possible for a
2528 request to fail to complete due to its worker thread's bad luck.
2529 When a request takes longer than the Windows SMB Redirector's timeout,
2530 the SMB Redirector tears down the SMB virtual circuit.
2532 When using the local directory updates it is really important that
2533 the directory update operations complete in the order that they were
2534 sent to the file server. If they don't, then the local directory state
2535 and the file server state will not match and the local directory state
2536 must be discarded which in turn forces a new read of the entire
2537 directory contents over the network.
2539 This patch adds a new cm_scache_waiter_t object that is used to store
2540 the current thread, buffer, and syncop flags within a waiters queue
2541 on each cm_scache_t object. If a thread is forced to sleep in
2542 cm_SyncOp, upon waking it will check to see if there are any other
2543 threads waiting that are attempting to perform a similar task ahead
2544 of it in the queue. If yes, the thread goes back to sleep. If not,
2545 it goes ahead and enters the cm_SyncOp conflict resolution block.
2547 This patch has the additional side effect of reducing the number of
2548 competing threads that must obtain the cm_scache_t mutex and process
2549 the cm_SyncOp conflict resolution block. As a result, the overall CPU
2550 utilization of the service and the clock time associated with processing
2551 requests will be reduced.
2553 * assert that the cm_scache_t mutex is held when calling
2554 buf_ForceDataVersion().
2556 * rename findNode to leafNode in bplus_Lookup
2558 * replace all OutputDebugString calls with osi_LogX calls
2560 * modify bestMatch to special case the return values for leaf nodes.
2561 If an entry is above or below the values available in the leaf node
2562 return BTLOWER or BTUPPER instead of BTERROR.
2564 * In insert and delete operations check for BTLOWER/BTUPPER and isleaf,
2565 if true convert to either slot 0 or Max and perform the insertion.
2566 This produces easier to read code when performing lookups.
2568 * modify lock_AssertXXX macros to call osi_assertx() and provide a
2569 descriptive message.
2571 * the difference between osi_assert() and osi_assertx() is whether or not
2572 a description is specified as part of the call. When no message is
2573 specified afsd_notifier() is called with a NULL msgp parameter. This
2574 results in a null pointer dereference during the EventLog() call.
2576 As a result, none of the information describing the assertion is written
2579 This commit sets a default message for use when no other message is
2580 provided and it replaces all calls to osi_assert with osi_assertx and
2583 * Migrate B+ tree search key into thread local storage
2586 * Fix an uninitialized variable in lana_IsLoopback() which
2587 could result in a random determination that a loopback
2588 adapter is not a loopback adapter. This would result in
2589 a computed Netbios service name of %MACHINE%-AFS instead
2590 of AFS. This is a serious problem if the value computed
2591 by the AFS Client Service and the client process pioctl
2592 calls produce different results as it will result in a
2593 communication failure.
2595 * Fix a crash in the B+ Tree code that results when a file
2596 rename only results in a change of case.
2598 * Fix the pioctl path parsing. The previous algorithm
2599 broke the current directory of the client into two parts,
2600 the TIDpath which represented the device root and the
2601 relative path. The problem is that this is not sufficient
2602 from proper processing of symlinks and mount points.
2604 What is required is that the last component of the path
2605 must not have the symlink or mount point be evaluated
2606 whereas symlink and mount point evaluation must be performed
2607 for the rest of the relative path. Therefore, the path is
2608 now split into three parts: TIDpath, intermediate path,
2611 "fs examine" now reports the data of the symlink or mount
2612 point object instead of the target and "symlink list" and
2613 "symlink remove" now works. The "Explorer Shell Extension"
2614 menu now is displayed for symlink objects.
2616 * More informative symlink.exe error messages.
2618 * When the cmd.exe "del" command is used to remove a symlink,
2619 not only was the link object removed, but the stat cache
2620 object for the target was marked deleted. This resulted
2621 in cache corruption which would require the destruction
2622 of the cache to correct.
2624 * Add undocumented registry value, "OfflineReadOnlyIsValid"
2625 which results in cached contents from read only volumes
2626 to be treated as valid when there is no callback registered
2627 and all servers are down or all instances are offline.
2629 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
2631 * Add documented registry value, "AcceptDottedPrincipalNames"
2632 which is used to permit the acquisition of AFS tokens
2633 using Kerberos v5 principal names which include a dot in
2634 the first component.
2636 {HKLM\HKCU}\Software\OpenAFS\Client
2638 * "fs uuid" does not require "AFS Client Administrator" group
2639 membership to display the current UUID value.
2642 * Update NSIS installer to version 2.30
2644 * Update Wix installer to version 2.0.5325
2646 * 64-bit KFW support:
2647 - Network Identity Manager plug-in
2649 - 64-bit krb5 enable afscreds.exe, afs server manager,
2652 * Enable B+ tree Directory Search by default.
2654 * NIM plug-in is now compatible with FILE: ccaches
2656 * Install instloop.exe into Client/Programs/ even though
2657 it isn't required. Permit it to be used by organizations
2658 that clone Windows systems.
2660 * On a suspend, only drop callbacks for servers that
2661 are online. If the server is offline, maintain the callback
2662 until the server is once again available.
2664 * Prevent against acquisition of a directory write lock when
2665 there are no updates to be performed. This was resulting
2666 in a write-lock leak which later on would lead to a deadlock.
2668 * Add configurable file pre-fetch mode. For files that match
2669 the specified file extensions, pre-fetch the entire file into
2672 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2673 (MULTI_SZ) PrefetchExecutableExtensions
2675 * Fix the client_osi large integer support so that return values
2676 can be properly assigned to large integer variables.
2678 * Do not hold cm_scache_t mutex across cm_GetCell() calls
2679 since cm_GetCell() can block in a RPC.
2681 * Add short file name support to B+ tree searches.
2683 * Resolve a conflict between the desire to notify file servers
2684 that the client workstation is suspending and the desire to
2685 permit objects with valid callbacks when the client suspends
2686 to continue to use those objects from the cache if the client
2687 resumes without network connectivity.
2689 * Fix stack variables to ensure that all buffers storing paths
2690 can handle AFSPATHMAX (1024) characters. Paths of 256 characters
2691 would cause a stack overflow condition.
2693 * On Vista, force afs_config.exe to run with the highest privileges
2694 available to the end user. For a user in the Administrators
2695 group running under UAC, this will cause afs_config.exe to be
2696 be run with full Administrators privileges after the user
2697 approves the privilege escalation.
2699 * More VNOVNODE handling issues. If a dirty buffer cannot be
2700 written due to VNOVNODE, clear the buffer and ignore the error.
2701 If a cm_scache_t object has been marked deleted due to a VNOVNODE
2702 error, the next time the smb_fid_t object that references it is
2703 used return either NOSUCHFILE or NOSUCHPATH and close the
2704 smb_fid_t object. Subsequent attempts to use the smb_fid_t will
2705 result in an invalid handle error.
2707 * For Vista, set the RequestExecutionLevel of the NSIS installer
2710 * For the build environment, support the generation of manifests
2711 when using the CL1310 (VS .NET 2003) compiler/linker.
2713 * There was a bug in OutputDebugHexString() which could overflow
2714 a stack buffer resulting in unexpected process termination
2715 of afsd_service.exe if the input Security Token contained
2716 many bytes equivalent to the '%' char.
2718 * Add directory lookup statistics to "fs memdump" output.
2722 * Windows uses case-insensitive file name pattern matching
2723 but AFS is a case sensitive file system. The AFS3 directory
2724 format is block based, uses network byte order and
2725 includes a hash table for fast case sensitive lookups.
2726 This causes several problems for the Windows AFS client.
2727 (1) Traversing the directory blocks is cpu expensive
2728 (2) A hash table miss does not indicate that the desired
2729 entry does not exist.
2730 (3) Determining whether a non-ambiguous inexact match or
2731 the entry does not exist requires a linear traversal
2732 of the entire directory.
2733 These issues often result in 100% CPU utilization.
2735 These issues are addressed by building a modified B+ tree for
2736 each directory and then using the B+ tree for searches.
2738 This feature is disabled by default. To enable it set
2739 the "BPlusTrees" registry value to 0x01 in the AFS Client
2740 Service Parameters key.
2742 * Correct a null pointer dereference in the symlink recursion
2743 detection code. (Introduced in 1.5.22)
2745 * Add the Network Identity Manager AFS Provider to AMD64
2750 * A new registry value, "BlockSize", can now be used to adjust
2751 the size of the internal buffers used to store file data.
2752 The default is 4KB. When set, the values 1..1024 specify
2753 multiples of 4KB. Values > 1024 specify actual block sizes
2754 and must be multiples of 4KB. The actual block size in KB
2755 can not be larger than the chunk size which defaults to 1MB.
2757 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2760 * pts.exe and ptserver.exe built with -DSUPERGROUPS
2762 * Service Manager shutdown hints have been added. The hint
2763 specifies that it may take the AFS client service up to two
2764 minutes to shutdown.
2766 * Fix the recursion detection feature introduced in 1.5.22.
2767 The previous implementation broke the evaluation of paths
2768 using ../path or ./path. It also broke the evaluation of
2769 symlinks to absolute paths. For example, /afs/cell/path.
2771 * Fix an uninitialized variable in cm_MakeDir that could result
2772 in crash if the requested directory could not be created.
2774 * Fix a the VolumeStatusNotification log messages to save
2775 the input string in the circular log buffer so garbarge
2776 won't be output to afsd.log when the log is dumped.
2779 * Increase the SMB Maximum Path length to 260 characters
2780 from 256 as that is what CIFS supports.
2782 * On 64-bit Windows, use a unique description string for
2783 the 32-bit AFS Shell Extension.
2787 * In the AFS Admin Library, replaced all of the
2788 Ubik_Call(XXX_RPC,...) calls with Ubik_XXX_RPC(...). This
2789 allows parameter checking and proper type conversions to be
2790 performed. This is crucial when passing parameters that are
2791 64-bit wide on systems in which 'long' is a 32-bit value.
2792 Otherwise, the parameter list on the stack is interpretted
2793 incorrectly and the function not only doesn't succeed but it
2794 might corrupt memory and if lucky crash the application.
2796 The only program that we care about that uses this library
2797 is the AFS Server Manager. Now it doesn't crash.
2799 * Fixed smb directory searches for non-wildcard names. The
2800 optimized function should not have set the bulk in progress
2801 flag. This would adversely affect later directory searches
2802 that required a bulk search.
2804 * Detect infinite recursions (up to 512 stat cache objects
2805 in a chain.) If an infinite recursion is detected, return
2806 an error to stop the CIFS client from querying us forever.
2808 * Executables executed out of AFS name space can have their
2809 pages swapped out by Windows. Those pages will be reloaded
2810 when they are required. However, if the cache manager no
2811 longer has a callback and all of the file servers with
2812 the volume data are down, then the application will crash
2813 due to the inability to read the required page.
2815 We now track the down time of each server. If the callback
2816 expires after the server that issued it went down and all
2817 servers are down, we delay the expiration to permit the data
2818 in the cache to be continue to be used by Windows or an
2821 * The Windows AFS client has always had very poor performance
2822 with regards to directory management. In the case of
2823 directory updates caused by create, rename,and delete operations,
2824 the AFS cache manager will now perform local directory updates
2825 when possible in order to avoid re-reading the directory from
2826 the file server. This will be done in whenever both of the
2827 following conditions are true:
2829 - the data version on the directory increased by one as a
2830 result of the operation
2832 - all of the directory buffers necessary to make the change
2833 are present in the cache
2835 * The error message in the AFS Control Panel describing the
2836 need to be a Windows Administrator has been replaced with
2837 a message describing the need to be in the AFS Admins Group.
2839 * If the vldb server list is empty when attempting to evaluate
2840 a cell mountpoint it was possible for a crash to occur. This
2841 can only happen if there is an existing mount point and the
2842 is a failure attempting to load the server list from the
2843 CellServDB file or via AFSDB queries.
2845 * dirty buffers only have their dirty bytes stored to the file
2846 server instead of the entire chunksize
2848 * the default chunksize has been increased from 128KB to 1MB
2853 * changed the enum values for cm_serverRef_t state info to use a
2854 private name space to avoid collisions (srv_)
2856 * added a srv_deleted state for cm_serverRef_t objects. This
2857 state is set when cm_FreeServerList() is called with the
2858 CM_FREESERVERLIST_DELETE flag set. cm_FreeServerList() may
2859 not always delete the cm_serverRef_t from the list if it is
2860 still in use by another thread. the srv_deleted state means
2861 the object's contents are no longer valid and it must be
2862 skipped. It will be deleted the next time the object is
2863 freed and the refcount hits zero.
2865 * the srv_deleted state is also used when a file server reports
2866 either VNOVOL or VMOVED instead of marking the cm_serverRef_t
2867 as offline. This is done to prevent additional usage of the
2868 stale vldb data while waiting for the update volume request
2871 * added a state field to the cm_volume_t object (enum volstate
2872 vl_ name space) that maintains the state of the volume based
2873 upon the states of all of the cm_serverRef_t and cm_server_t
2876 * modified cm_UpdateVolume() to set the state of the cm_volume_t
2877 RW, RO, and BK to either vl_alldown or vl_online. There can't
2878 be any other states because cm_UpdateVolume() destroys any
2879 previous knowledge we might have had regarding busy or offline
2882 * modified cm_UpdateVolume() to update the volume name in the
2883 cm_volume_t to the volume base name if the previous value was
2886 * modified cm_FollowMountPoint() to check to see if the volume
2887 name is a volume ID and if so call cm_GetVolumeByID instead
2888 of cm_GetVolumeByName. This ensures that volume IDs are always
2889 looked up as numeric values. There is no longer a need to
2890 maintain a separate cm_volume_t containing the string representation
2893 * Added a flags parameter to cm_GetVolumeByName() and cm_GetVolumeByID().
2894 The first flag is a "CREATE" flag which is set by all existing
2895 calls. The flag is not set by calls to cm_GetVolumeByID() from
2896 the server probe code when volume status is being updated. We
2897 do not want the server probe operation to result in additional
2898 turnover in the cached volume data. The second flag is NO_LRU_UPDATE
2899 which is set when the server probe code updates the volume status.
2900 This flag will be used to prevent the server probe operation from
2901 changing the order of the least recently used queue.
2903 * Modified cm_GetVolumeByName to ensure that only one cm_volume_t is
2904 allocated for a given set of normal, readonly, and backup volumes
2905 regardless of whether or not the volume is accessed via name or
2906 ID number. The cm_volume_t namep field is always the base name
2909 * Added a new volume state, vl_unknown. This state is used as
2910 the initial state for all cm_volume_t when the cache manager starts,
2911 for each cm_volume_t at creation, and for each cm_volume_t when
2912 recycling. The cache manager does not know the state of all
2913 volumes in the world, only those that are in the cache and for
2914 which it has queried the VLDB and hosting file servers.
2916 * modified cm_GetVolumeByName() to initialize the state of a
2917 volume to vl_unknown. The actual state will be set when a
2918 cm_VolumeUpdate() call completes successfully.
2920 * changed name of scache hash table variables to avoid ambiguity
2921 when adding hash tables for volumes
2923 * fix a buffer overrun in sys\pioctl_nt.c pioctl().
2925 * modified cm_UpdateVolume() to handle the case in which there is
2926 no RW volume but there is are RO volumes for a given base name.
2927 This is done by querying for the ".readonly" volume name if the
2928 base name does not exist in the VLDB. We never query for the
2929 .backup name because under the current usage model a .backup
2930 volume may only exist on the server that the read-write volume
2931 is located. If there is no RW volume, there can be no .backup.
2933 * Added four hash tables for cm_volume_t objects to improve the
2934 search time of cm_GetVolumeByID(), cm_GetVolumeByName() and
2935 cm_ForceUpdateVolume(). One each for Name, RWID, ROID, and
2936 BKID. Three ID hash tables are necessary as long as it is
2937 desireable to maintain a single cm_volume_t containing all
2938 of the related RW, RO, and BK volume data. Having the RW and
2939 RO volume data in the same object is necessary for the
2940 implementation of cm_GetROVolumeID() which returns either the
2941 RO or RW ID depending upon the existence of RO volume instances.
2943 * Added a volume LRU queue so that volume reuse becomes fairer.
2944 This does not replace the all Volumes list which is used when
2945 it is desireable to walk a list of all the volumes whose order
2946 is not going to change out from underneath you which makes it
2947 safe to drop the cm_volumeLock.
2949 * handles volume hash table updates where volume name to
2950 volume ID number changes. The volume name remains
2951 constant in the cm_volume_t. if a vos rename is performed,
2952 the name of the volume will change and the volume IDs will be
2953 updated. Subsequent access to the old volume ID will create a
2954 new cm_volume_t with the new name.
2956 * Added a daemon thread operation to query the state of volumes
2957 listed as busy or offline. cm_CheckBusyVolumes() calls
2958 RXAFS_GetVolumeStatus() for each volume ID that is marked vl_busy
2959 or vl_offline. If the volume is now online, the status on the
2960 volume is updated. The default period is 600 seconds. This can
2961 be configured with the BusyVolumeCheckInterval registry value.
2963 * Added prototype for smb_IoctlPrepareRead() which was missing a
2964 return type in the function definition.
2966 * Added volume id lists to the cm_server_t. These lists are
2967 allocated in blocks of ~32 IDs. When a cm_PingServer()
2968 detects a change in server state, the state of the cm_volume_t
2971 * Added volID to the cm_serverRef_t object. volID is used
2972 to identify the volume for which the object is a referral.
2973 cm_FreeServerList() uses the volID to remove the volume
2974 from the cm_server_t.
2976 * In cm_Analyze, when VNOVOL or VMOVED are received,
2977 call cm_ForceVolumeUpdate() to force a refresh of the volume
2980 * Added cm_VolumeStatusNotification() which is used at the moment
2981 to log volume status changes to the trace log. It will also
2982 be used as the access point to the File System Filter driver
2983 notification engine.
2985 * Added an all cm_scache_t list to cm_data. This replaces the use
2986 of the stat cache LRU queue when we need to enumerate all
2987 entries. The LRU list order is not static and when using it to
2988 enumerate all entries it can result in items being missed or
2989 items being processed more than once.
2991 * Modified cm_Analyze(). Instead of reseting the busy or offline
2992 state of a volume and forcing a retry of the operation
2993 cm_Analyze will defer to the background daemon thread that will
2994 update the state once every 600 seconds.
2996 * Added the automatic generation of a Freelance ".root" read-write
2997 mountpoint that refers to the root.afs volume of the workstation
2998 cellname at the time the mountpoint is created.
3000 * re-write cm_Analyze to make better use of the known volume
3001 status. VL_Server queries cannot result in CM_ERROR_ALLOFFLINE
3004 * renamed cm_CheckBusyVolumes to cm_CheckOfflineVolumes.
3005 busy volumes will be reset to srv_non_busy by the function
3006 but there is no mechanism for querying the busy state other
3007 than by attempting to access the resource.
3009 * cm_Analyze will query the state of an offline volume before
3010 deciding whether or not to retry when all volume instances
3013 * the algorithm used after an IP address list change detection
3014 has been modified. we now check the state of the VLDB servers,
3015 then reset the rx connections, and finally check the state of
3016 the file servers. Previously we flushed the rx connections
3017 and then checked the servers in whatever random order they
3018 were first discovered.
3020 * fix the volume status reported by "fs examine"
3022 * various improvements to the AFS Network Identity Manager Provider.
3023 - Don't access freed objects after the window has been destroyed.
3024 - Write back the AFS cell list in the identity configuration
3025 even when its empty.
3026 - Disable all use of the provider when AFS is disabled for a
3027 particular identity.
3029 * Fix cm_cell_t leak. When evaluating a mountpoint that uses an
3030 alias instead of the full cellname, double check the full cell
3031 name and see if we have a matching cell before allocating a new
3034 * Dump more data with "fs memdump". All volumes and scaches and
3037 * Reverse the order of the all cells list. This ensures that
3038 the workstation cell is always first in the list.
3040 * If we recognize the server from which a callback revoke was
3041 received, then we can ensure that we only invalidate the status
3042 information for cm_scache_t objects belonging to that cell.
3044 * When building for amd64, use APPVER 5.02 as we do not support
3045 64-bit Windows 2000 and explicitly state the machine type in
3048 * no longer treat VMWare network adapters as equivalent to a
3049 loopback adapter. For some unknown reason, publishing the "AFS"
3050 netbios name on them frequently results in a name collision
3053 * Use read locks whenever possible instead of write locks when
3054 walking the allCellsp list. This permits better use of multiple
3057 * In 1.5.20 a bug was introduced by the fake status when PRSFS_READ
3058 is not present on the directory optimization. It would report
3059 mount points and directory objects for which there was no
3060 status information as a zero length file. Instead we can use
3061 the vnode number of the object to determine if it should be
3062 considered to be a directory or a file. If it is odd it is a
3063 directory; even it is a file.
3065 * Give Back Callbacks to the file servers in response to suspend
3066 and shutdown power management events. Only give up callbacks
3067 to UP servers and do so with a very short RX dead timeout.
3070 * Fix a race condition in the SMB FID cleanup code that could
3071 result in a stat cache mutex never being released.
3073 * Fix the graphics files that are installed as part of the
3074 IBM Administrators Guide.
3076 * Change the behavior of pioctl path parsing to not follow
3077 symlinks or mount points during evaluation. This results in
3078 the FID being the symlink or mount point and not the object
3079 they point to. This is consistent with the behavior of the
3082 * Enable AFSDB record lookup for command line tools such as
3085 * CellServDB update 2007-07-05
3087 * Added a Volume Status Plug-in Interface that permits third parties
3088 to receive notifications when the volume state changes occur.
3090 * Ensure that all writes to the afsd_alloc.log file use CR-LF.
3092 * Prevent the afsprocmgmt.dll from being unloaded while the
3093 background thread it starts is running.
3095 * Avoid a race condition when closing smb_fid_t objects.
3098 * fix install path of IBM Quick Start Guide when installed by
3101 * fix the client so that it can write into directories with
3102 only 'li' permissions.
3104 * fix installation of 32-bit Shell Extension for 64-bit systems.
3105 Install it under its own HKCR keys.
3107 * modify the wix installer to remove and upgrade install packages
3108 with the same version number.
3110 * when flushing volume location data every two hours or in
3111 response to "fs checkvolumes", also flush the mount point
3112 links in order to force their re-evaluation.
3114 * aklog was copying an uninitialized buffer as part of the
3115 foreign cell automatic pts registration
3117 * when obtaining directory listings, check for read permission
3118 on the directory. If read permission does not exist, fake the
3119 status information for the directory entries. This improves
3120 performance, prevents the client from pounding the file server,
3121 and prevents leakage of cached status information.
3123 * "fs getcacheparms" output was incorrect.
3125 * NetIdMgr AFS Provider should only automatically add token
3126 acquisition for cell foo.com if the realm of the cell matches
3127 the realm of the identity.
3129 * Fix a null pointer dereference in cm_MergeStatus()
3133 * network identity manager provider:
3135 - Correct for the fact that pr_CreateUser tampers with the
3136 contents of the username input buffer.
3138 - Take into account that the krb4 credential data structure
3139 does not have separate realm fields for the user and service.
3140 When krb524 is used with a cross-realm ticket, the realm
3141 describes the service not the user.
3143 * Many integrated logon changes:
3145 - Fix eventlog reporting. Do not attempt to log an event
3146 if the event source registration fails. Use DebugEvent0
3147 instead of DebugEvent when there are no parameters.
3149 - Modify the LOOKUPKEYCHAIN macro to recognize ERROR_MORE_DATA
3152 - Fix the reading of Domain specific configuration for
3153 LogonScript and TheseCells. Previously the dwSize value was
3154 being overwritten so that subsequent RegQueryValueEx call
3157 - Fix a memory leak in the TheseCells reading code.
3159 - Add support for Domain specific "Realm" specification. The
3160 realm is the realm to be appended to the username. When
3161 logging in as a domain or to the local machine, the specified
3162 "Domain" name is not going to be a valid realm name.
3164 - Construct a proper principal name based upon the domain
3165 specified realm for use in obtaining tokens with KFW.
3167 - If the domain specified "TheseCells" list includes the
3168 default cell, do not obtain tokens twice.
3170 * An unprotected call to the Kerberos v5 function,
3171 krb5_free_context(), can result in an invalid memory access
3172 exception if the contents of the krb5 profile, krb5.ini, cannot
3173 be parsed correctly. The error is located within the afskfw
3174 library which is used by afscreds.exe, afssvrmgr.exe, and
3175 afslogon.dll. When the error is triggered within afslogon.dll
3176 a blue screen condition can occur when the winlogon.exe system
3177 process terminates unexpectedly.
3181 * Do not stop the afs client service with an assert
3182 if the cell name cannot be determined when responding
3183 to a "fs whereis" command.
3185 * Integrated Logon (afslogon.dll). No longer attempt
3186 to preserve Kerberos 5 tickets for the logon session
3187 as that job is now supported by KFW 3.1 and above.
3189 * The Network Identity Manager plug-in will attempt to
3190 remove expired service tickets from the cache.
3194 * CellServDB update - GCO Public CellServDB 09 Mar 2007
3196 * AFS Server Manager and AFS Account Manager data cache
3197 moved from HKCU\Software\TransarcCorporation to
3198 HKCU\Software\OpenAFS to prevent crashes due to data
3201 * Reduce the default UP server probe time from 10 minutes
3202 to 5 minutes to match UNIX. This will not solve the NAT
3203 port mapping timeout issue because most NATs default to
3204 a value less that 5 minutes and this value ensures that
3205 the afs client server will start probes 5 minutes after
3206 the last probe set ended. The probe period for any one
3207 server is therefore longer than 5 minutes. The more
3208 servers in the list the longer the time between probes.
3210 * Before marking a server as down in cm_Analyze, retry the
3211 request using a new rx connection to the server. If the
3212 cache manager is behind a NAT, it is possible that the
3213 response from the server is being sent to the original
3214 address:port from which the rx connection was established.
3215 This will appear to the cache manager as the server being
3216 down when in fact it is not. Forcing the use of a new rx
3217 connection will result in the server responding to the
3218 most recent address:port assigned to the client by the NAT.
3220 If the server is in fact down, it will simply take a little
3221 longer to notice it.
3225 * Removing a directory that contained entries was failing with
3226 the wrong error code being returned to the CIFS client.
3230 * rxdebug exponential back-off. don't take down a server that
3231 is already struggling to respond.
3233 * Network Identity Manager plug-in registers help on the Help
3234 menu if it is supported by NIM. (KFW 3.2 and above)
3236 * Plug-n-play networking did not always work after coming back
3237 from a Windows Suspend. The AFS client service failed to
3238 re-compute the LAN adapter to bind to and instead attempted
3239 to bind to all LAN adapters. This would polute the public
3240 network with the name "AFS" which could lead to conflicts
3243 * Build installers using Microsoft Hotfixes provided at
3244 http://tinyurl.com/243svk for Visual Studio C RunTime Library
3245 Daylight Savings Time issues.
3247 * "fs checkserver" does not require admin privileges unless
3248 the probe interval is being changed.
3252 * A reference leak was discovered in the mountpoint traversal
3253 functionality which resulted in cm_volume_t objects not becoming
3254 eligible for re-use. This will trigger a panic condition when
3255 more than the maximum number of volumes is accessed.
3257 * Fixed the success test for the "fs memdump" operation. Success
3258 is now reported as such. Added cm_volume_t object information
3261 * A data corruption error has been fixed. Write requests
3262 queued for background processing were not retried upon failure.
3263 This leads to inconsistencies between the contents of the file
3264 in the local cache and the contents of the file on the file
3267 The implemented fix is to re-queue any request that has failed
3268 due to timeout, volume busy, volume offline, server busy, or
3269 server offline. Requests are not processed while all servers
3270 for a volume are inaccessible. When servers become accessible
3271 the background request will be retried.
3273 In addition, if a background write fails, do not mark the buffers
3274 as clean. Failure to preserve the dirty flag prevents the buffers
3275 from being written to the file server in the future.
3277 This fix does not address the possibility that the credentials
3278 necessary to perform a write request might expire between the
3279 time the request is queued and when the request can be processed
3280 due to server accessibility.
3282 * The Freelance volume does not support per user ACLs. The
3283 cm_HaveAccessRights function did not handle this special case.
3284 Requests for privileges greater than read|lookup would result
3285 in afsd_service.exe entering an infinite loop. ISOBuster is
3286 known to cause this behavior.
3288 * When deleting files, the SMB Server would attempt to obtain
3289 the access permissions for the parent directory instead of
3290 for the file itself. This would result in an incorrect
3291 assessment of whether or not the file can be deleted.
3293 * The Network Identity Manager AFS plug-in has had its string
3294 tables updated for consistency with the language used by NIM.
3295 All credentials are referred to as credentials instead of tokens.
3296 Drop-down combo boxes now in fact drop down.
3298 * The AFS Salvager was not properly being built on Windows.
3299 This is relevant only if the AFS servers are in use.
3301 * The default signal handlers for the AFS servers incorrectly
3302 processed SIGQUIT. Instead of stopping the server, an
3303 exception was raised.
3305 * Fix another volume refcount leak. This one was triggered by
3306 an inability to contact the vlservers in order to update the
3309 * Fix checklist control used by the drive mappings tabs in
3310 afscreds.exe and afs_control.exe.
3312 * Improve output of "fs memdump" related to buffer management.
3314 * Fix refcount leak of cm_buf_t objects when prefetching buffers
3315 already in the process of being fetched.
3317 * Improve cache performance by computing the buffer hash table
3318 size based upon the number of buffers so that the average
3319 number of entries in a bucket is seven.
3321 * Improve cache performance of read/write data by locally
3322 updating the dataVersion of all buffers associated with the
3323 file on each StoreData or StoreStatus operation provided that
3324 the dataVersion was incremented by one and the dataVersion of
3325 the buffer matched the original dataVersion of the cached
3326 status information for the file.
3328 * When recycling cm_scache_t objects, make sure that we
3329 clear all of the previous fields. It is no longer the
3330 case that we can assume that the stat cache object is being
3331 obtained only after a successful FetchStatus call.
3332 InlineBulkStatus will allocate stat cache objects for which
3333 the no status information is available.
3335 * When a file is being opened with the OPEN_ALWAYS disposition,
3336 do not require write permissions when the file already exists
3337 unless of course the file is being opened requesting write
3340 * When a file has been opened for read-only access, if the
3341 application requests an exclusive lock, obtain a read-lock
3342 instead of a write-lock. The purpose of the exclusive lock
3343 in this situation is to prevent changes to the file and the
3344 full file read lock will provide this guarrantee.
3349 * Enforce Unix Mode Write bit to prevent file deletion of read only
3352 * The CIFS READ and LIST permissions use the same value 0x0001.
3353 When a directory is being accessed, check for PRSFS_LOOKUP
3354 instead of PRSFS_READ.
3358 * Fix computations of short names of mount points. They were
3359 broken as part of the directory search optimization. (1.5.13a)
3361 * When all buffers are in use, periodically sleep the threads
3362 waiting for a buffer to become free while the global buffer
3363 lock is not being held in order to permit those threads that
3364 can free buffers to do so.
3366 * Power Management improvements. Maintain a global flag that
3367 specifies whether or not the service is in a suspend state.
3368 Do not panic if Netbios() returns NRC_BRIDGE meaning that the
3369 lana is no longer valid. Instead, stop the listener threads
3370 and if all listener threads are stopped, reset the lana_list.
3371 Allow the cm_Daemon() thread to periodically check the state
3372 of the smb listeners. If they are all stopped and the service
3373 is not suspended, attempt to restart them. If there are no valid
3374 lanas, return to the stopped state.
3376 * CreateX and NTCreateX use cm_CheckNTOpen() to test whether or not
3377 the user is permitted to obtain read or write locks. This function
3378 would obtain the lock and then drop it returning whether or not
3379 the lock could be obtained. If the lock was in fact required,
3380 CreateX/NTCreateX would then obtain it with cm_Lock(). The problem
3381 of course being that this pattern results in three RPCs to the
3382 file server (lock, unlock, lock). This is reduced to one RPC by
3383 implementing cm_CheckNTOpenDone() which frees the allocated byte
3384 range lock from cm_CheckNTOpen() after the cm_Lock() call is
3387 * Optimize the SMB FindFirst processing for the case in which there
3388 are no wildcards. If a cm_Lookup succeeds, generate a response
3389 from the stat cache entry we have a callback for. Its cheaper
3390 than walking each buffer of directory data associated with the
3391 directory stat cache entry when we have thousands of directory
3394 * Increase the SMB Ioctl MaxData size to support larger token sizes
3396 * Disable autorun of the AFS Server Wizard when the AFS Server is
3399 * Add AFS Server service dependencies on tcpip and PNP_TDI
3401 * Remove the AFS Server Configuration Tool shortcut
3403 * Add Freelance mode awareness to AFS Server Configuration Tool
3407 * Unlike previous versions of the OS, Vista performs a shutdown on
3408 the Microsoft Loopback adapter just like it would on a real adapter.
3409 This causes the smb_Listener threads to trigger a panic during a
3410 suspend/hibernate power event.
3412 The fix is to unbind from the network adapters in response to a
3413 suspend/hibernate power event and then rebind to the adapters
3414 when a resume power event is received. Note that the resume events
3415 are not reliably delivered so it is possible the afs service will
3416 not be accessible. However, this is the best we can do.
3418 * don't allow environment variables we set in afslogon.dll to be
3419 inherited by child processes
3421 * force an rx connection reset if VICECONNBAD or VICETOKENDEAD are
3424 * The old dirty buffer synchronization algorithm had a buf_IncrSyncer
3425 thread walking the all buffer list periodically searching for dirty
3426 buffers to write to the file server. This had several negative
3427 results. The algorithm ate up ever increasing amounts of CPU time
3428 even when AFS is idle as the size of the cache increases. Also,
3429 buffers were written to the file server in an order based upon the
3430 original buffer allocation which has nothing to do with the order
3431 in which the buffers became dirty.
3433 The new algorithm maintains a dirty buffer list. Items are added
3434 when the buffer is originally marked dirty. A buffer is only
3435 removed from the list by the buf_IncrSyncer when the buffer is no
3436 longer dirty. If the list is empty the thread goes back to sleep
3437 immediately without additional processing requirements.
3441 * Enable renaming of directories from the Explorer Shell when
3442 Offline Folders is enabled
3444 * An error is reported in response to the SMB ReceiveCoreFlush
3445 request when an error has not occurred.
3449 * optimize dirty buffer flushes in response to suspend operations
3450 do not flush if only loopback adapter is installed
3451 do no retry dirty buffer writes
3453 * do not reset netbios when performing loopback checks. doing
3454 so will break the existing lana bind
3456 * remove a deadlock in smb3.c QPathInfo
3460 * Add NetIDMgr AFS Plug-in
3462 * "Show Tray Icon" checkbox removed from afs_config.exe
3464 * Disable DNS Registrations for the Loopback Adapter and make sure
3465 that Netbios is turned on.
3467 * A reference count leak on stat cache objects was fixed
3469 * For debugging, tread ids of writers are now tracked on locks
3471 * Corrected an incorrect lock that was being held but not
3472 freed in a rare error case
3474 * Changed trace log end of line to CR-LF so it can be read
3477 * modify the buffer synchronization thread to write all dirty
3478 buffers and sleep only when there is no work to be done
3480 * improve behavior of cache manager when large numbers of
3481 temporary files are being created by multiple clients within
3484 * do not permit dirty buffers that belong to deleted files
3485 to be written to the file server. its a waste of bandwidth
3487 * do not leak stat cache references when the file server
3490 * afscreds: do not permit an attempt to obtain tokens if the
3491 afs service could not be started. this was causing stack
3492 overflows on Vista after resuming from a suspend
3496 * better handling of dot directory
3498 * fs sq and fs mkm now behave as on UNIX
3500 * add license to installers
3502 * Return Invalid Parameter error for unknown InfoLevels in
3503 SMB Tran2 Search Directory.
3505 * Correctly use WSAStartup and WSACleanup within the RX library
3507 * Check that afsd_service.exe is running before performing
3510 * Force AFS Shell Extension to load resource DLL using
3511 afs_shl_ext.dll as a base name.
3513 * (Wix) install afs_shl_ext_XXXX.dll and afs_creds_XXXX.dll in
3514 OpenAFS\Client\Program
3516 * Fix SMB_SET_FILE_END_OF_FILE_LENGTH
3518 * Do not panic if the maximum number of volume entries are in use
3519 and one of them can be recycled.
3523 * Fix the code that reads the backconnnectionhostnames value
3524 so that there is not an uninitialized variable
3528 * integrated logon could cause winlogin.exe to blue screen windows
3529 if it was called with a domain that does not exist on the machine.
3530 this is possible during a sysprep.
3532 * corrections to cm_buf_t handling to ensure that buffers are
3533 removed from queues before they are freed and that holds are
3534 obtained on the next buffer in a chain before releasing the
3537 * do not generate MessageBox dialogs on network errors. they
3538 cause more trouble then they are worth.
3540 * move GlobalAutoMapper into a separate thread so that it does
3541 not prevent service startup if the WNetAddConnection2 call
3544 * restrict hard dead and connection timeouts to the CIFS Session
3545 Timeout value. This is being done to try to prevent virtual
3546 circuits from being destroyed prematurely.
3548 * Fix the code that writes the backconnectionhostnames value
3549 to ensure that the data buffer is written with the correct
3552 (there were no windows specific changes between 1.5.4 and 1.5.5)
3556 * Corrections to the Integrated Logon functionality that permits
3557 Kerberos 5 TGTs to be communicated into the user session. The
3558 transfer is now more secure and succeeds when the user does not
3559 have read permission to the %SystemRoot%\TEMP directory.
3561 * NSIS installer updated to version 2.18
3563 * WiX installer updated to version 2.0.4221.0
3565 * Updates to the IFS functionality for 64-bit correctness and
3566 permit compilation. The IFS functionality works for pioctl()
3567 operations but not much else at this point.
3569 * A major re-write of the OpenAFS implementation of the CIFS Tran2
3570 query and set functions.
3572 * Ignore the Windows CIFS client session timeout entirely. Timeout
3573 only after the AFS Hard Dead Timeout period expires.
3577 * move headers, libraries, and samples from OpenAFS\Client to
3580 * remove the auto-registration of AFS IDs for foreign cells
3581 from integrated logon. this permits afslogon.dll to avoid
3582 initializing the RX library. afslogon.dll is a network
3583 provider library and as such it is loaded into all applications
3584 which perform a CreateProcessAsUser call. Initializing the
3585 RX library spawns background threads to listen for inbound
3586 packets. When the network provider is unloaded from the
3587 process, the RX library does not cleanup the threads. If they
3588 continue to execute after the library is unloaded, the process
3591 * implement inline bulk status calls. this will significantly
3592 improve the performance of directory listings.
3594 * prevent a crash when evaluating mountpoints to volumes that do not
3597 * new command: fs uuid [-generate]
3598 displays the current uuid and allows a new one to be generated
3599 on the fly without restarting the afs client service.
3601 * fix the 64-bit installers to properly document the install
3604 * move help files to the proper location so they can be located by
3605 the system tray tool and the control panel.
3607 * "over quota" and "disk full" errors were not being reported to the
3610 * largefile support has been added. Files greater than 31-bits can no
3611 longer be written to file servers which do not support large files.
3612 When file servers support large files, files up to 64-bits can be
3615 * cm_SyncOp could return with the input buffer unlocked upon error
3616 which would result in a panic condition. This bug was introduced
3619 * vlserver failover when freelance mode did not work. This prevented
3620 the root.afs volume from being loaded if the first vlserver could
3625 * A bug in the CIFS Standard Info response packet caused directories
3626 to appear as standard files. This prevented Microsoft Vista from
3627 being able to walk the AFS directory tree.
3629 * Mount points must be reported as directories in response to
3630 Query Path Info operations.
3632 * A deadlock was detected when performing "fs flushall" if the
3633 file server reports VNOVNODE. The scp->createBufferLock is
3634 already held by the current thread and there is no mechanism
3635 to propagate the knowledge. Therefore, an alternate mechanism
3636 for clearing the cache must be developed. A new function
3637 cm_RecycleSCache(scp,flags) has been extracted from
3638 cm_GetNewSCache(). This function performs the task of recycling
3639 an cm_scache_t object. When called from cm_GetNewSCache() with
3640 no flags the expectation is that there are no associated buffers
3641 that are queued to be read or written. When called from
3642 cm_Analyze() with the CM_SCACHE_RECYCLEFLAG_DESTROY_BUFFERS
3643 flag, any queued buffers will be de-queued and marked as if
3644 the operations were performed so the data can be discarded.
3646 * CM_ERROR codes were logged as "unknown" by cm_Analyze.
3648 * In response to VNOVNODE, the parent is only discarded if the current
3649 cm_scache_t is not a directory.
3651 * In the Ioctl FlushFile and FlushVolume functions, there were no
3652 checks to protect against flushing the Freelance SCache entries.
3654 * In FlushFile, the wrong cm_scache_t object was being released.
3656 * In cm_GetNewSCache, do not allow Freelance SCache entries to be
3657 recycled. Choose a new entry if cm_RecycleSCache fails.
3659 * Speed up the performance of the cache manager by not
3660 holding the smb_fid_t mutex across calls to cm_SyncOp
3663 * Ensure that all smb_fid_t flag references are protected.
3665 * Remove a deadlock between smb_fid_t mutex and
3666 smb_rctLock introduced in 1.4.1.
3668 * Fix aklog AFSID lookup to use not send the realm
3669 name when the realm is the local realm for the
3670 cell. The PT Server doesn't strip the realm
3671 name when it is the local realm.
3673 * Treat "\\afs\*." as an alias for "\\afs\all"
3675 * Dynamically adjust the priority of server threads
3676 based upon the age of the CIFS request that is
3677 being processed. The priority is bumped one level
3678 for every 15 seconds of processing time.
3680 * The original openafs contribution mis-used the
3681 cm_GetCallback function. This function should only
3682 be called as a side effect of a call to
3683 cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK). The way it
3684 was being called results in a FetchStatus RPC being
3685 performed even when the client already has a valid
3686 callback and does so in a manner that destroys the
3687 synchronization of the threads that are calling
3688 cm_SyncOp. If nothing else this results in a
3689 significant performance penalty.
3691 * More changes to cleanup of smb_vc_t objects in
3692 order to prevent race conditions.
3694 * Reduce the amount of Power Management event logging
3697 * If the client detects that the IP address has
3698 changed, force the use of new RX connections. This
3699 enables the file server to recognize that the client
3702 * Remove a race condition and deadlock associated
3703 with cleanup of smb_vc_t objects.
3705 * Fix the ANSI filename support. One function that
3706 required it was missing the translations.
3708 Changes not present in 1.4.1 and since 1.5.0: 1.5.1 released
3711 * Add a DllMain() function to afspthread.dll. Perform
3712 cleanup of thread local storage, rx queues, critical
3713 sections, event handles, etc. This was implemented
3714 to avoid crashes in afspthread.dll when applications
3715 which load/unload the dll attempt to terminate.
3717 * Lock request behavior has been altered. As of this
3718 release a locally allocated lock will be used whenever
3719 the volume is RO or the requesting user's permissions
3720 are 'rl'. This will allow end users to execute programs
3721 from RW volumes when they do not have the 'k' privilege.
3723 Note that a recent discovery indicates that the 'k' bit
3724 is only supposed to control the ability to obtain read
3725 locks. Write locks are to be controlled by the 'w' and
3728 * Code has been added to enable the cache manager to be
3729 able to obtain the AFSID of the authenticated user.
3730 This is required to be able to determine when "Insert"
3731 privileges should apply. AFS considers a file to be
3732 a new file whenever the owner's AFSID matches the
3733 AFSID of the user. The current implementation relies
3734 on the name specified during the SetToken call to be
3735 correct. This will be replaced in the future with a
3736 new RPC that allows the AFSID of the authenticated
3737 user to be obtained.
3739 * A 32-bit-tools installer for use on 64-bit systems
3740 is now available. These tools are for use with 32-bit
3741 versions of Kerberos for Windows or JPSoftware's 4NT.
3743 Since 1.4.1 rc10: 1.4.1 final released (9 April 2006)
3745 * Remove the use of the "active_vcp" variable. This variable
3746 no longer serves any useful purpose and was the cause of
3747 race conditions that produced under counts of the smb_vc_t
3750 * Change default minidump output to include data segments
3752 * restore access to cm_IoctlGetVolumeStatus and
3753 cm_IoctlSetVolumeStatus calls for AFS pioctl operations.
3754 Still need to figure out why they are being called
3755 for SMB/CIFS IOCTL operations instead of Trans2GetFileInfo.
3757 Changes not present in 1.4.1: 1.5.0 released (16 Feb 2006)
3759 * Update build system to include the "sys_name" in the
3762 * optimize the number of calls to obtain the smb_user_t for a
3763 smb virtual circuit.
3765 * add logging of server addresses, status and capabilities when
3768 * prevent multiple threads from pinging the same server simultaneously.
3769 instead, threads block and then use the result of the ping that was
3770 already taking place.
3772 * Add registry values to allow the defaults for daemon checks to be altered
3773 at startup. Log values to afsd_init.log.
3775 daemonCheckDownInterval
3776 daemonCheckUpInterval
3777 daemonCheckVolInterval
3778 daemonCheckCBInterval
3779 daemonCheckLockInterval
3780 daemonCheckTokenInterval
3782 * Add EventLog messages for transitions in fs crypt state.
3784 * Byte range locking (when locks will be obtained from the
3785 is still under discussion)
3787 * 64-bit Windows operating systems now supported
3788 (XP SP2 64-bit, 2003 SP1 64-bit, 2003 R2 64-bit)
3789 This is a native 64-bit implementation of OpenAFS for Windows.
3790 There is currently no 64-bit version of KFW and therefore
3791 there is no (easy) method for obtaining AFS credentials.
3792 32-bit KFW 3.0 can be used with 32-bit OAFW libraries
3793 to obtain AFS tokens but there is not for the feint of
3796 * A faster implementation of the fcrypt algorithms improves
3799 * The afsd_service.exe now supplies Event Log Message Text.
3800 The event log no longer generates warning text due to lack
3801 of message configuration.
3803 * When afs servers are unreachable and network stack returns
3804 a WSAEHOSTUNREACHABLE error, the AFS client will immediately
3805 mark the attempted server as being down. This allows for
3806 very fast response to the calling application. This change
3807 has a pronounced improvement when the Windows Explorer is
3808 used on disconnected mobile computers.
3810 * Added DebugOutputString debugging of the RX library
3812 Since 1.4.1 rc9: 1.4.1 rc10 released (10 March 2006)
3814 * Fix the rx-lwp implementation to randomly generate rx call
3815 identifiers which are supposed to be unique at a given point
3816 in time. This affects some of the command line tools used
3817 for debugging but not the AFS Client Service itself which
3818 uses the rx-pthreads implementation.
3820 * xstats_fs_test.exe and xstats_cm_test.exe are now built for
3821 Windows although they are not distributed
3823 * don't generate an exception when logging unexpected SMB packets
3825 * the data structures used by cm_IoctlGetVolumeStatus and
3826 cm_IoctlSetVolumeStatus are not compatible with the CIFS
3827 Get/Set File Information calls so don't bother executing them.
3831 * Fix data written to registry as part of BackConnectionHostnames
3834 * Fix initialization bug when allocating cm_server_t objects that
3835 would cause servers to appear to be down after each allocation
3836 or renewal. Server would appear up after the next check servers
3841 * Windows will return WSAECONNRESET to the next receive packet
3842 operation on a socket if an ICMP packet is received in response
3843 to a packet that was sent and could not be delivered because
3844 the port was unreachable. We have to treat this as
3845 WSAEWOULDBLOCK so that we can retry the select() check and
3846 not mark the connection as bad.
3848 Since 1.4.1 rc6: 1.4.1 rc7 released (14 Feb 2006)
3850 * The client manager maintains a least recently used list of
3851 all the ACLs. If an ACL is already the most recently used
3852 do not bother to remove it from the list and place it on the
3855 * As part of the process of adding a new server, set the
3856 server is down flag and then perform a ping. If the server
3857 is up, the flag will be reset. If the server is down, the
3858 client will not timeout unnecessarily while attempting to
3861 * If a Mountpoint String is the empty string, return PATH NOT
3864 * Modify behavior of the Network Provider to avoid querying the
3865 profile location if integrated logon is known to be disabled.
3867 * Correct behavior of 'LogoffPreserveTokens'. When set, tokens
3868 will always be preserved regardless of whether or not integrated
3869 logon was used to obtain tokens. When unset, tokens will be
3870 destroyed regardless of whether integrated logon was used to
3871 obtain the tokens unless the profile is stored in AFS.
3873 * Disable LogoffTransferToken functionality. The SMB V3 Logoff
3874 message is being delivered even when the user is not logging
3875 off of the computer.
3877 * Fix a deadlock condition introduced in rc6.
3879 * Cell names are case-insensitive. Since the founding of OpenAFS
3880 there has been a bug that treated "cell", "Cell", and "CELL" as
3881 different cells. This would result in what appeared to be
3882 random disappearance of user tokens or a failure to authenticate
3883 even when tokens were listed by the "tokens" command.
3885 Since 1.4.1 rc5: 1.4.1 rc6 released (2 Feb 2006)
3887 * Additional auditing of the smb code revealed that smb ioctl
3888 operations were failing to free smb_fid_t objects
3890 * The File Identifiers assigned to smb_fid_t objects could be
3891 given the value 65535 which is 0xFFFF which is also the
3892 INVALID_HANDLE value. Both 0 and 65535 are now treated as
3895 * Audited all assignments to fields and ensured that they are
3896 protected by obtaining the relevant mutex object.
3898 * Re-wrote the VNOVNODE handling to avoid a potential deadlock
3899 if both the current cm_scache_t object and its parent are
3900 both returning VNOVNODE.
3902 * Removed all the logic associated with the 'dead_vcp' variable
3903 which must have been added for debugging because its used
3904 was never thread safe and it truly served no purpose.
3906 Since 1.4.1 rc4: 1.4.1 rc5 released (25 Jan 2006)
3908 * Re-enabled LogoffTokenTransfer and LogoffTokenTransferTimeout
3909 registry settings. Default is ON and 120 seconds. If set to
3910 OFF, tokens will not be destroyed.
3912 * Fixed a number of race conditions and reference count problems
3913 associated with global queues.
3915 * Added a method to detect abandoned smb virtual circuits so their
3916 associated file handles, locks, user tokens, and smb session
3917 resources can be released.
3919 * Added a mechanism to allow user tokens to be preserved
3920 during integrated login and be freed after logoff.
3922 * Check access permissions to the file before the file is
3925 * More improvements to the handling of SMB Virtual Circuits
3926 when SMB sessions are closed.
3928 * Correct a race condition introduced in rc4 that could result
3929 in a crash of afsd_service.exe.
3931 Since 1.4.1 rc3: 1.4.1 rc4 released (11 Jan 2006)
3933 * Properly handle pre-mature SMB session closures.
3934 File handles, tree connections, and user sessions are now
3935 freed when the virtual circuit is closed.
3937 * smb_username_t and smb_user_t objects are now properly reference
3938 counted and freed as appropriate.
3940 * disable to use of AFS file locks. byte range locking is still
3941 implemented but all locks are managed locally. This prevents
3942 multiple processes from editing the same file on the same
3943 machine but it does prevent multiple processes on separate
3944 machines from modifying the file.
3946 Since 1.4.1 rc2: 1.4.1 RC3 released (4 Jan 2006)
3948 * A fix to "fs setserverprefs -vlserver". Multiple calls with
3949 the same server parameter would result in a panic condition
3950 if the server had not already been contacted by the AFS
3953 * Better warnings are written to the afsd.log file when
3954 the Windows RPC Protocols are not properly configured as
3955 is often the case on Windows 2000.
3957 * fix the procmgmt library to restore C RTL signal handlers when
3958 it is being detached from the process prior to process termination.
3960 Since 1.4.1 rc1: 1.4.1 RC2 released (30 Nov 2005)
3962 * close a lock leak in the rx library that could be triggered while
3965 * prevent removal of drive mappings in response to afscreds.exe -M
3967 * add logic to process VNOVNODE in cm_Analyze. Force re-evaluation of symlink
3968 strings and flush the stat cache entry.
3970 * prevent server objects from being freed if user preferences are set.
3972 * Update WIX support to 2.0.2419.0
3974 * Release all locks on the server at service shutdown.
3976 * A down server should be marked up as long as it responds to the client.
3977 It doesn't matter if the server returned an error as long as the error
3978 is something other than VRESTARTING. If there is such an error, all
3979 connections on that server should be marked for replacement.
3981 * Change the check "up" servers period from once per hour to once every
3982 ten minutes. This matches the behavior of the Unix cache manager.
3984 Post 1.4.0 revisions: 1.4.1 RC1 released (21 Oct 2005)
3985 * Byte Range Locking support has been added. Architectural documentation
3986 is located within src/WINNT/afsd/cm_vnodeops.c. Best efforts are
3987 made to enforce Windows mandatory locking model using locally managed
3988 byte range locks with full file locks obtained from the file server.
3989 Read-only volumes provide the user with an implied locking privilege.
3990 Read-write volumes require that a user be granted the locking privilege
3991 in order for a lock to be obtained. As most Windows applications use
3992 Shared file access modes, they require the ability to obtain locks in
3993 order to access files.
3995 * Implement error translation functions to convert Universal Error Codes
3996 output by modern servers to local errno and winsock error values.
3997 This fixes the error message problems using pioctl() calls from fs.exe
3998 and the AFS Shell Extension.
4000 Since 1.4.0 rc8: 1.4.0 final tagged (20 Oct 2005); released (31 Oct 2005)
4001 * Allow the AFS Client Service to set the firewall configuration by
4002 retrying until after Windows determines the boot period is over.
4004 * Do not attempt to configure the firewall on Windows 2000, XP pre-SP2,
4007 * Detect IP address changes from within the AFS Client Service for the
4008 purpose of checking down servers sooner
4010 * If the file server or vldb server report RXKADUNKNOWNKEY, return
4011 SEC_E_NO_KERB_KEY to Windows. This provides an indication to the
4012 user that there is a Kerberos key problem. Previously the user would
4013 receive no feedback when there was something wrong.
4015 * Fix the large integer library support to properly handle values greater
4018 Since 1.4.0 rc7: 1.4.0 rc8 released (15 Oct 2005)
4019 * Updated CellServDB
4021 * Fixed file name pattern matching. Question marks can match nothing
4022 if found at the end of a component.
4024 Since 1.4.0 rc6: 1.4.0 rc7 released (8 Oct 2005)
4025 * Kerberos 5 ticket lifetimes should adhere to KFW Leash lifetimes
4027 * The MSI installer now has a new Product ID for each build. This
4028 is to allow for automated upgrades. The Upgrade ID is remaining
4031 * Volumes that were cached and then subsequently were salvaged,
4032 moved or otherwise taken temporarily offline (as opposed to being
4033 busy) would become unreachable. A distinction is now made between
4034 a volume being offline and a server being down.
4036 Since 1.4.0 rc5: 1.4.0 rc6 released (2 Oct 2005)
4038 * Fixed several AFS Shell Extension issues:
4039 - Checking Server Status would only work if a specific cell was
4041 - If servers were down, the Server Down dialog would write to
4042 invalid memory and cause explorer.exe to crash
4043 - Default Push Buttons were not specified for many dialogs
4044 - The extension would be unloaded by the Explorer Shell and
4045 never be reloaded under some circumstances.
4047 * Fixed the AFS System Tray menu to disappear if another Window
4048 was selected before a menu item was selected
4050 * Added volume owner and group information to "fs examine" output
4052 * Added fs and registry support for enabling/disabling/clearing
4053 rx statistics gathering.
4055 Since 1.4.0 rc4: 1.4.0 rc5 released (25 Sep 2005)
4057 * New Product Code GUIDs issues for 1.4 and 1.5 branches. Separate
4058 GUIDs for each platform.
4060 * Documentation directory structure and default installation rules
4063 * New OpenAFS for Windows HTML Release Notes installed
4064 as part of the documentation set.
4066 * Add thread id to rx debugging messages
4068 * When all servers are offline, return a bad network path error
4069 immediately. This is necessary to allow Explorer to do the
4070 right thing and not hang. The background daemon thread will
4071 check the down servers every 30 seconds to see if they are
4074 Since 1.4.0 rc3: 1.4.0 rc4 released (14 Sep 2005)
4076 * The AFSCache file has been modified to store both the
4077 serial number of the volume on which the file resides
4078 and the SID of the local machine. If either change,
4079 a new UUID is generated.
4081 * add "fs flushall" command that forces all data buffers
4084 Since 1.4.0 rc2: 1.4.0 rc3 released (4 Sep 2005)
4086 * add support for '/' instead of '\' in pioctl() calls
4088 * Apply AFS Client Admins group protect to AFS Shell Extension
4090 * Add support for \\afs\<name> to most AFS Shell Extension
4091 commands except the symlink methods.
4093 * when installing the Microsoft Loopback Adapter, enable
4094 MS Client for Networks and prevent an install failure
4095 by not calling CoInitialize twice in the same thread.
4097 * reload cell vldb values from the CellServDB every two
4098 hours in case it changes
4100 * When updating cell information from DNS, be sure to set
4103 * Add support to allow use of \\AFS\<foo> where <foo> is
4104 either a mount point or symlink. As <foo> is normally
4105 treated as a share name, we transform it into \\AFS\all\<foo>
4108 * Init 'code' to prevent false errors when integrated logon
4109 is disabled and the service is not running
4111 Since 1.4.0 rc1: 1.4.0 rc2 released (28 Aug 2005)
4113 * Removed trace log messages that were hampering performance.
4115 * Fixed a deadlock that was being triggered by editing Word
4116 documents stored within AFS with WinWord 2003.
4118 * Bit 3 of the TraceOption registry value is now used to set the
4119 default for "fs trace" in the non-Debug builds. The new default
4120 is off for release builds and on for debug builds.
4122 * Bit 2 of the TraceOption registry value can be used to turn on
4123 real-time output of debug log entries to the Windows Debug
4124 Output monitor interface. This data can be viewed with tools
4125 such as Sysinternal's DbgView.
4127 Since 1.3.87: 1.4.0 rc1 released (17 Aug 2005)
4128 * When the cache manager reports ALLBUSY or ALLOFFLINE for
4129 a resource we should not be returning status codes such
4130 as Network Busy or Remote Host Down because these errors
4131 will imply that the AFS Client Service is Busy or Down
4132 and that is simply not the case. Instead we will return
4133 Bad Network Path as the path is temporarily not available.
4135 Instead of returning Sharing Paused when there is a
4136 timeout error, return Timeout. Once again, the AFS Client
4137 Service is not paused.
4139 * afscreds.exe would display an Obtain Creds dialog when
4140 the expired credentials reminder was triggered even
4141 if there was no network path to the KDC. This is prevented
4142 by adding KDC probe logic to the reminder thread.
4144 * afscreds.exe would display expired tokens no differently
4145 than unexpired ones. This would make it difficult for the
4146 user to distiguish when the tokens were expired. For the
4147 English build added a new resource string "(expired) that
4148 is displayed instead of the expiration time.
4151 * "fs wscell" when executed in freelance mode will return the
4152 name of the cell configured in the registry. The root.afs
4153 volume is not loaded from this cell, but it is used for the
4154 default for aklog and integrated login.
4156 * "fs mkmount, fs rmmount, symlink make, symlink remove" will
4157 no longer work on \\AFS\all when freelance is being used
4158 unless the user is a member of the "AFS Client Admins" Windows
4161 * some more checks performed during persistent cache validation.
4164 * Optimize calls to cm_CheckServer from cm_Analyze. Only check
4165 down servers for the current cell not all cells.
4167 * If the scache LRU list becomes corrupted, the AFS Client Service
4168 might enter an infinite loop at startup. Detect the corruption
4169 and discard the cache contents in this case.
4171 * Fixed the Explorer Shell's remove mountpoint functionality.
4173 * Fixed a deadlock caused by the holder of cm_aclLock attempting to
4174 obtain a mutex lock on a cm_scache_t object whose lock is held by
4175 a thread attempting to obtain the cm_aclLock.
4178 * Really fixed DNS AFSDB queries to ensure that "csail" cannot be
4179 misinterpretted as "csail.mit.edu" when the resolver
4180 is configured to append ".mit.edu" to failed DNS queries.
4182 * Added a new registry key, "LogoffPreserveTokens" (see registry.txt),
4183 that can be used to force the preservation of user tokens upon logout.
4185 * Update the NSIS install scripts to use NSIS 2.07. This release adds
4186 recommended installation categories: AFS Client, AFS Administrator,
4187 AFS Server, AFS Development Kit. Each category includes a different
4188 default subset of the OpenAFS components.
4190 The OpenAFS logo is now associated with the NSIS uninstall entry
4191 in the Add/Remove Programs control panel.
4193 * The user name associated with AFS tokens when obtained with
4194 integrated login, the afs systray tool, or aklog will always
4195 include the full Kerberos 5 user principal name regardless of
4196 whether or not the cell is local to the realm.
4198 * Modify integrated login so that it does not enter an infinite
4199 loop if the service is not set to auto start.
4201 * Added asetkey.exe used to set a Kerberos 5 key for use by the
4204 * Added uninstall.exe to Wix installation
4206 * More modifications to algorithms used to wake sleeping threads.
4210 * Changes to the algorithms used to wake threads when they are
4211 put to sleep because of conflicts over buffer operations.
4214 * The OpenAFS integrated logon obtains Kerberos 5 tickets as part of
4215 the process of producing AFS tokens. The tokens are stored within
4216 the AFS Client Service but the Kerberos 5 tickets have been discarded.
4217 New functionality has been added to temporarily copy the tickets to
4218 a file ccache during the short window between the NPLogonNotify and
4219 WinLogon Logon Event. The Logon Event handler starts a process as
4220 user, afscpcc.exe, that copies the tickets into the default ccache.
4221 The file ccache is then destroyed by the event handler.
4223 * During synchronization operations on stat cache entries, if threads
4224 are forced to sleep it was possible that the threads would not be
4225 woken under heavy load.
4227 * Enforce the delete on close flag specified by the NTCreateX()
4230 * A race condition in the rx library was fixed that resulted in
4231 a deadlock in rx_NewCall when the RX_CONN_MAKECALL_WAITING flag
4232 was cleared when threads were still waiting to use the
4233 connection. Also, fixed a potential case in which the
4234 connection could be destroyed while threads were waiting to use it.
4236 * The DNLC freelist has been seen to become corrupted with still
4237 active entries being stored on the freelist. Changes were made
4238 to perform a better job of cleaning entries before placing them
4239 onto the freelist; marking them with the correct magic value when
4240 purging the list; and allowing the list to be purged on startup
4241 if the cache validation fails.
4243 * Windows Crash Reporting does not get triggered for afsd_service.exe
4244 because it provides its own top level exception handler. This patch
4245 places an explicit call to ReportCrash() on platforms which support
4248 If you configure Crash Reporting via AD Policy to capture crash
4249 reports within the domain, then you will receive this output.
4251 * Generate MiniDumps independent of the Windows Crash Reporting.
4253 (a) If an exception occurs,a minidump will be generated at the path
4254 %TEMP%\afsd.dmp. The type of minidump defaults to Normal but
4255 can be set to any other type via use of the
4258 [HKLM\SOFTWARE\OpenAFS\Client] MiniDumpType (DWORD)
4260 (b) MiniDumps can be generated at any time using the "fs minidump"
4261 command. This allows you to generate a minidump without
4262 stopping the process or even requiring the use of a debugger
4265 * an RX connection which reports bad ticket now treats it the same way
4266 the expired ticket is treated. The ticket is removed from the
4267 connection and a new anonymous connection is established.
4269 * fs memdump now requires that the caller be in the AFS Client Admins
4272 * additional debugging information is provided in cm_Analyze when the
4273 error code is ignored.
4275 * Fix a regression introduced into the pioctl() client code which
4276 will cause a crash in winlogon.exe, explorer.exe and fs.exe if
4277 MIT KFW is not installed. (1.3.82a)
4279 * Remove AFS Gateway option from OpenAFS Control Panel (1.3.82a)
4282 * Fix a race condition in cm_EndCallbackGrantingCall() that
4283 could leave 'nrevp' pointing to freed memory. In the process,
4284 optimize the processing to ensure that the applications monitoring
4285 the status of the stat cache entry are only notified at most
4288 * Conditionalize access to the Explorer Shell AFS->Mount Point
4289 ->Remove menu item. Disable the item if the selected item(s)
4290 are not mount points.
4292 * Activate AFS RPC (RX) Free Packet Queues stored in Thread
4293 Local Storage. This eliminates thread contention on the
4294 global RX free packet queue. This should improve performance
4295 on multi-CPU systems.
4297 * Fix the IP address filtering code to properly load/unload
4298 the IPHelper DLL so that Windows XP/2003 and beyond systems
4299 do not have to manually probe the registry.
4301 * CleanupACLEnt() was not being called consistently with
4302 the cm_scache_t object referred to by the back pointer
4303 mutex locked. This could in very rare conditions lead to
4304 an invalid memory access.
4306 * Added a script command to the msi installer "afs_replace"
4307 which can be used to replace one version of OpenAFS with
4308 another without requiring an uninstall and reinstall if
4309 the installed msi is no longer in its original location.
4311 * Apparently the problem with multi-domain forests with cross-
4312 realm trusts to non-Windows realms was not entirely solved.
4313 The authentication to the AFS SMB service failed because
4314 the wrong name was being used. Using ASU as an example,
4315 the authentication was being performed with the name
4316 "QAAD\user" (an account in the forest root) and not
4317 "user@ASU.EDU (the MIT Kerberos principal used to login with)
4319 The solution was to add an additional dependency on KFW
4320 in order or to be able to easily obtain the client principal
4321 name stored in the MSLSA ccache TGT. This information is
4322 used in two locations:
4324 - the pioctl() function
4326 - a new WinLogon Event Handler for the "logon" event.
4328 The pioctl function will now be able to use the correct
4329 name when calling WNetAddConnection2() and the "logon"
4330 event handler will now be able to call WNetAddConnection2().
4331 The hope is that the "logon" event handler will be called
4332 before the profile is loaded but I have not guarrantee
4336 * Fixed a locking error in cm_TryBulkStat() which had the
4337 potential of deadlocking the system for the length of time
4338 it takes to perform a bulk status fetch operation.
4340 * Replaced time conversion code (UnixTime <-> FILETIME) to
4341 be completely arithmetic instead of relying on a bizarre
4342 algorithm involving a variety of C RTL time functions.
4343 This has the side effect that UnixTime and FILETIME which
4344 are both stored in UTC are interpretted as UTC throughout
4345 the year. Windows will apply the same localization to AFS
4346 as it does to NTFS. Applications which rely on the ability
4347 to sync files between the two file systems will no longer
4348 see the timestamps of files in AFS change an hour relative
4349 to the files stored in NTFS or Windows based backup devices.
4351 * Fixed a invalid memory access under a bizarre circumstance.
4352 Windows will allow a physical mass media device to be
4353 installed via Plug N Play to the system and will assign it
4354 the lowest available drive letter. This is true even if the
4355 drive letter is currently assigned in the user session to
4356 a network device via NET USE (or its equivalent.) When this
4357 happens, queries sent to the CIFS server will contain invalid
4358 data. This invalid data was not being caught by the AFS
4359 Client Service and was resulting in a crash when Freelance
4362 * Fixed a reference count error when registering callbacks
4363 on a stat cache object if the callback was already registered
4365 * Add a case to cm_Analyze() in order to handle the case where
4366 cm_GetConnByMServers() returns CM_ERROR_NOSUCHVOLUME because
4367 the server list for the volume is empty. In this case, force
4368 an update of the volume info and retry.
4370 * Insert a missing cm_EndCallbackGrantingCall() which could
4371 result in threads waiting for a callback to complete to never
4374 * In the persistent cache, there is a maximum number of volume
4375 entries. Allow volume entries to be reused if the maximum
4376 number have been allocated and their reference count is zero.
4378 * If we already have a dead virtual connection object, don't
4379 ignore the fact that additional dead vc objects must be taken
4382 * Removed a deadlock condition introduced in the summer of 2004.
4383 Do not hold mx locks around calls to RX functions including
4384 rx_NewCall(). That is what reference counts on the rx_connection
4387 * Fixed an initialization error in afslogon.dll which could
4388 result in random behavior including a failure to terminate
4389 the NPLogon function. (1.3.80b)
4391 * Fixed an error preventing the use of SMB authentication
4392 on Windows 2000. (1.3.80a)
4395 * Updated CellServDB to Public CellServDB 16 Mar 2005
4397 * Fixed DNS AFSDB queries to ensure that "csail" cannot be
4398 misinterpretted as "csail.mit.edu" when the resolver
4399 is configured to append ".mit.edu" to failed DNS queries.
4401 * Fixed another case in which the client would replace connections
4402 to the host when it was not necessary. In this case the
4403 situation would occur if cryptall was on and the connections
4404 were unauthenticated due to lack of an appropriate token.
4406 * OpenAFS for Windows has failed to work at sites which are
4407 utilizing a cross-realm trust between an MIT/Heimdal realm
4408 and a multi-domain Windows forest when the workstation being
4409 accessed is not located in the root domain. This is caused
4410 by a bug in the workstation which was triggered after the
4411 introduction of Windows 2003 Server. When the bug is triggered,
4412 the workstation attempts to authenticate users to afsd_service.exe
4413 by contacting the Domain Controller instead of using the
4414 LSA loopback authentication mechanism.
4416 One of the reasons this bug occurs is because the workstation
4417 does not have a reliable way of knowing that the service whose
4418 netbios name is "AFS" is located on the workstation. This will
4419 be fixed starting in Longhorn Beta 1 by Microsoft. The
4420 "BackConnectionHostNames" registry value will be used to
4421 indicate that the authentications to that service name should
4422 be performed using the loopback authentication mechanism.
4424 In the meantime, when Logon Caching is enabled, we can force
4425 afsd_service.exe to authenticate using the logon cache before
4426 contacting the Domain Controller. This will work with both
4427 password and smart card based logons.
4429 * The allDown logic in cm_ConnByMServers() was wrong. The allDown
4430 flag should not be cleared if a volume's server reference is
4431 marked as "offline". In the case where all of the volume's
4432 servers are either "down" or the volumes are "offline", we want
4433 cm_Analyze() to process the condition as CM_ERROR_ALLOFFLINE
4434 instead of as CM_ERROR_TIMEDOUT. In fact, CM_ERROR_TIMEDOUT
4435 should never occur in practice.
4437 In the case of CM_ERROR_ALLOFFLINE, cm_Analyze() will sleep for
4438 5 seconds, clear the server down and volume busy flags, and
4439 then force an update from the VLDB. This allows the client to
4440 update the location of a volume if the reason for it being
4441 marked offline is because it is being moved. Calls to
4442 cm_ConnByMServers() will be retried either until success or
4443 the RDRtimeout period is reached.
4445 * Correct the Power Management code behavior in response to
4446 Standby, Suspend, and Shutdown events. Instead of flushing the
4447 buffers associated with the mounted SMB submounts, simply write
4448 all buffers which are marked dirty.
4450 * Added support for "TheseCells" to afslogon.dll. "TheseCells"
4451 provides a list of cells other than the default cell for which
4452 tokens should be obtained using the default Kerberos principal.
4453 This functionality is logon domain specific and is only available
4454 when using KFW for authentication.
4456 * Fixed FindNext Invalid Handle error caused by over agressive
4457 attempts at garbage collecting dirSearch entries when the
4458 dirSearch ID wraps from the maximum value to 1.
4460 * Add support for registry defined server preferences for VLDB and
4461 File servers. See registry.txt for details.
4463 * Increased default cache size to 96MB and default number of cache
4466 * Fixed refCount leaks related to directory lookups and pioctl calls.
4468 * Callbacks revoked during a race condition with an attempt to
4469 obtain the same callback no longer result in an inaccessible
4472 * New command line tool:
4474 afsdacl : Set or reset the DACL to allow starting or stopping
4475 the afsd service by any ordinary user.
4477 Usage : afsdacl [-set | -reset] [-show]
4478 -set : Sets the DACL
4479 -reset : Reset the DACL
4480 -show : Show current DACL (SDSF)
4482 * IP addresses are no longer queried once at startup. Instead IP
4483 addresses are obtained as needed. Loopback adapter addresses are
4484 no longer published to the server.
4486 * Pay attention to the MIT KFW registry configuration for automatic
4487 importation of MSLSA credentials. (SOFTWARE\MIT\Leash32,MsLsaImport)
4489 * Fix (once again) case-sensitive comparisons which was apparently
4492 * Activate support for vos listvol -format
4494 * Population of the "AFS Client Admins" group failed on non-English
4495 versions of Windows because the name of the "Administrators" group
4496 is localized. Now we lookup the name of the group by using the
4497 Administrators Group Alias Relative ID.
4499 * Multiple cell token acquisition within afscreds.exe was broken in the
4500 case of Kerberos 5 cross realm authentication. Instead of contacting
4501 the KDC associated with the cell's realm, afscreds would obtain a
4502 token from the KDC of the user's realm. This would result in
4503 "invalid kvno errors" while authenticating to the AFS servers.
4504 Unauthenticated access would work. The symptoms would vary based upon
4505 whether or not the VLDB servers had been contacted using unauthenticated
4506 connections prior to the user obtaining tokens.
4508 * The list of ACL entries was becoming corrupted. It appears as if
4509 an ACL when it expires was not being appropriately cleaned up. In
4510 fact, it was left in the list of ACLs associated with the scache
4513 * Changed the default @sys name list to "x86_win32 i386_w2k i386_nt40"
4514 for 32-bit x86 systems. The default for itanium will be "ia64_win64"
4515 and "amd64_win64" for amd 64-bit processors.
4517 * When executing executables, Windows provides a mechanisms by which
4518 the normal search for DLLs can be bypassed. If a file foo.exe is
4519 being executed and there exists a file or directory "foo.exe.local"
4520 (the contents are ignored if a file), then Windows will search for
4521 DLLs first in the "foo.exe.local" directory and second in the
4522 directory in which the "foo.exe" file is located.
4524 Previous releases of OAFW would improperly return
4525 STATUS_NOT_A_DIRECTORY instead of the expected
4526 STATUS_OBJECT_PATH_NOT_FOUND. This would cause the Windows SMB client
4527 to terminate the search for the DLL causing the execution of the
4528 application to fail.
4530 * Rework the reference counting of smb_vc_t objects. The references
4531 stored in the waiting locks were not counted. This could result in
4532 an assertion if the reference count drops to zero.
4534 * "fs wscell" returns "Freelance.Local.Root" for the workstation cell
4535 when running in freelance mode.
4537 * Added support for persistent caching of file contents, stat entries,
4538 volume data, cell data, and name lookup data. When the registry
4539 value, "NonPersistentCaching", is not defined or set to 0 the file
4540 specified by "CachePath" becomes a persistent cache file. The
4541 size of the cache file is computed at run time. It has a maximum
4542 size of approximately 1.2GB. The contents of the file will be
4543 validated according to the rules specified by the "ValidateCache"
4546 * Added support for UUIDs. UUIDs are kept for as long as the
4547 cache file is intact.
4549 * Added cmdebug.exe and support cache manager debugging callback
4550 interfaces in afsd_service.exe.
4552 Usage: cmdebug -servers <server machine> [-port <IP port>] [-long]
4553 [-addrs] [-cache] [-help]
4554 Where: -long print all info
4555 -addrs print only host interfaces
4556 -cache print only cache configuration
4558 * Symlinks to \\AFS\[all\]... will now be treated the same as
4559 symlinks to /afs/.... However, please use /afs/... as otherwise
4560 the symlinks won't work on Unix.
4562 * Correct a problem with local CellID allocation for cells whose
4563 server list is obtained via DNS instead of a CellServDB file. If
4564 the DNS information expires the CellID assigned to the entry will
4565 be changed. This causes all of the cm_scache_t objects which refer
4566 to the old cellID number to become useless. Attempts to access
4567 files or directories with cm_scache_t objects using the old cellID
4568 will fail since the server list cannot be obtained.
4570 * Correct deadlock condition in cm_EndCallbackGrantingCall which can
4571 be triggered if the volume referenced by the cm_scache_t object is
4574 * The AFS Service needs to respond to SERVICE_ACCEPT_SHUTDOWN messages
4575 in addition to SERVICE_ACCEPT_STOP.
4577 * Move RPC shutdown until after the SMB and RX shutdown procedures
4578 complete. Block until RPC shutdown is complete.
4580 * Modify afslogon.dll (integrated logon) to wait for service if its
4581 state is START_PENDING. If the timeout period occurs, reset to
4582 the retry interval and not the sleep interval.
4584 * When renewing the server list for a cell obtained via DNS AFSDB
4585 records, the cm_cell_t entry must be removed from the list of all
4586 cells. Otherwise, the list of cells will be corrupted.
4588 * In the dcache and scache modules, use the cm_scache_t dataVersion
4589 instead of the cm_fakeDirVersion.
4591 * Synchronize fs.c with the unix version.
4593 * The variable used to determine whether a file or virtual memory
4594 mapped cache is used was not properly initialized to a default
4595 value. If the registry setting "NonPersistentCaching" was not
4596 set, the choice would be random. Properly initialized to be
4599 * The memory mapped view was never unmapped before closing the file
4600 at service shutdown. This is now properly cleaned up.
4602 * Default location of Cache file is now %TEMP%\AFSCache
4606 * A bug affecting new installations of 1.3.75/76 would result in
4607 the creating of incorrect mountpoints in the freelance root.afs
4608 volume for the default cell.
4610 If "fs \\afs\all\<cellname>" lists a volume name of "root" instead
4611 of "root.cell", you have been affected by the bug. To correct the
4612 problem, execute the following commands:
4614 fs rmmount \\afs\all\<cellname>
4615 fs rmmount \\afs\all\.<cellname>
4617 fs mkmount \\afs\all\<cellname> root.cell <cellname>
4618 fs mkmount \\afs\all\.<cellname> root.cell <cellname> -rw
4622 * A bug has been fixed in the auto-generation of Freelance root.afs
4623 symlinks which produced random entries in the \\AFS\all directory.
4625 * Support has been added for multi-homed servers
4628 * Added a new registry value, "StoreAnsiFilenames", which can be used
4629 to force the use of ANSI character sets instead of OEM Code Pages.
4630 This feature is useful when users require the ability to create
4631 filenames with 8-bit characters and need to access the files from
4632 both Latin-1 based Unix systems as well as from Windows.
4634 Activation of this feature will prevent access to files stored with
4635 8-bit OEM characters.
4637 * Shutdown all SMB threads in a synchronized manner when stopping the
4640 * There is currently a maximum cache size of 1.3GB. The limit is imposed
4641 by the largest contiguous block of unused memory within the 2GB process
4642 space which can be assigned to the memory mapped file. Unfortunately,
4643 when the executable digital signature verification code is activated
4644 Windows sees fit to further segment the process memory which in turn
4645 reduces the size of the maximum cache file to less then 800MB. If
4646 larger cache sizes are desired, a new registry value should be set:
4648 HKLM\SOFTWARE\OpenAFS\Client (DWORD) "VerifyServiceSignature" = 0x0
4650 Setting this value will disable the runtime verification of digital
4651 signatures on afsd_service.exe and the afs dlls which it loads. It
4652 will not disable the the version number check on those same files.
4653 The signature verification is not a security messure and is only meant
4654 to enhance the ability to afsd_service.exe to detect potential
4655 destablizing mixtures of DLLs from incompatible distributions.
4657 Added code to auto-disable the signature verification check if
4658 the desired cache size is greater then 700MB.
4660 * Windows' WinTrustVerify(WIN_SPUB_ACTION_PUBLISHED_SOFTWARE) is
4661 used to verify the validity of the afsd_service.exe binary
4662 as well as each of the AFS DLLs loaded by the service. Not only
4663 must the digital signature be valid but the signatures of the
4664 DLL must be signed by the same entity as the service.
4666 * Implement new functions: cm_freelanceMountPointExists and
4667 cm_freelanceSymlinkExists. Use them along with other validity
4668 checks in cm_freelanceAddMount and cm_freelanceAddSymlink to
4669 ensure that name collisions do not occur and that empty strings
4670 are not valid file names.
4672 A symlink may not have a name which would resolve to a valid
4673 cell name. Doing so would prevent access to the cell.
4675 * Add missing cm_HoldSCacheNoLock call to Freelance mount point
4676 re-initialization code. The reference counts of the fake root.afs
4677 volume scache object(s) would become invalid when the mount point
4678 or symlink lists were altered.
4680 * Add registry entries to provide mappings from the afsdsbmt.ini
4681 to the new locations for applications which count on the use
4682 of the old Profile file APIs. These apps are likely to fail
4683 if the user does not have administrator privileges and the
4684 registry is locked down.
4686 * The afs_config.exe submounts dialog had two errors.
4687 First, attempts to remove entries failed because the registry
4688 key was being opened without KEY_WRITE privileges.
4689 Second, when editing a submount entry, changing the name
4690 would add a new key and leave the original one in place.
4691 Now the original submount will be removed if its name is
4694 * In recent months there have been several incidents in which
4695 users have experienced problems starting or accessing
4696 afsd_service.exe and after significant effort has been spent
4697 it has turned out that they have two versions of AFS on the
4698 machine or an inconsistent set of DLLs.
4700 Code has now been added to afsd_service.exe which will walk
4701 the list of modules loaded by afsd_service.exe and validate
4702 that the version of the AFS DLLs matches the version of the
4703 afsd_service.exe executable. If they do not match the service
4706 * When Freelance mode is enabled and there is no registry
4707 key HKLM\SOFTWARE\OpenAFS\Client\Freelance, afsd_service.exe
4708 will attempt to import the afs_freelance.ini file contents.
4709 If the file does not exist, it was creating a dummy file
4710 with a r/o and r/w entry for the default cell and then
4711 importing those values.
4713 This process has been changed. The temporary file is no
4714 longer created. Also, both the OpenAFS Client install
4715 directory as well as %WINDIR% are checked for previous
4716 afs_freelance.ini files.
4718 * Added support for VL_GetEntryByNameN(). Still need to add
4719 support for VL_GetEntryByNameU() for multi-homed support.
4721 * Fix a deadlock situation in afscreds.exe when canceling an
4722 auto-generated Obtain Tokens dialog
4725 * Fix the locking of objects during Directory Searches in the
4726 SMB/CIFS server. The failure to properly lock the reference
4727 counts was resulting in the premature freeing of smb_dirSearch_t
4728 objects while they were still in use by the SMB/CIFS client.
4729 This does not solve the "Invalid Handle" problem.
4731 * Fix Find Cell By Name pioctl call to return a valid cell
4732 name for the Freelance fake root.afs volume.
4733 "Freelance.Local.Root".
4735 * Fix the Explorer Shell Extension Symlinks->Add operation.
4736 The dialog template was missing and the link destination
4737 string was too short.
4739 * Add support for symlinks to Freelance root.afs volume
4740 Stored at HKLM\SOFTWARE\OpenAFS\Client\Freelance\Symlinks
4741 <number> = "<linkname>:<relative-path>."
4742 Use symlink.exe to create, list, or remove
4744 * Remove the fallback to the use of KFW's KRB4 library when obtaining
4745 tokens. We never obtain KRB4 tickets.
4747 * Fix AFS Client Configuration Control Panel to support new SysName
4750 * Fix a bug in afsd_service.exe which could result in the SysName
4751 not being read from the registry.
4754 * Add code to block the issuance of AFS tokens by aklog.exe or
4755 afscreds.exe when the Kerberos 5 principal name contains a dot.
4757 * Modify the IsAdmin() function to always treat the local SYSTEM
4758 account as an AFS client administrator. Affects fs.exe and
4761 * Modify the internal handling of Quota Exceeded errors
4763 * Upgrade all reference count fields in the Windows cache manager
4764 and the osi library to use unsigned long instead of signed short.
4765 A similar fix has been applied to the afs rpc (rx) library.
4767 * fix the Windows cache manager to prevent it from replacing the
4768 rx_connection object associated with the cm_conn_t object on each
4769 and every operation if "fs crypt" was set. This explains the
4770 dramatic performance difference when crypt is used vs clear.
4771 The problem: 'cryptall', a boolean flag indicating whether or not
4772 "fs crypt" is set, was being compared to the rx_connection
4773 cryptlevel which is either rxkad_clear:0 or rxkad_crypt:2.
4774 1 != 2 and therefore the rx_connection was always destroyed
4775 and replaced on each and every operation.
4777 Lock the cm_conn_t object around every call to RXAFS_xxxx functions.
4778 It is not safe for the cm_conn_t object to not be locked because
4779 rx_DestroyConnection might be called from another thread if:
4780 - the user's tokens have changed or expired
4781 - the crypt mode has changed
4783 This fix appears to have also taken care of the problems associated
4784 with Overlapped Writes resulting in Delayed Write errors.
4786 * fix NSIS installer's AdminGroup.exe to properly create and
4787 remove groups when given -create or -remove. The string comparison
4790 * fs sysname now accepts a list of sysname values
4792 * added a new registry value HKLM\SOFTWARE\OpenAFS\Client "IoctlDebug"
4793 DWORD which when set to a non-zero value will cause error message
4794 text to be output to stderr from the pioctl() routine. Useful in
4795 debugging failures of fs.exe, tokens.exe, etc.
4797 * added a test to the power management code to only perform a
4798 flush operation if there is at least one network adapter which
4799 is not a loopback adapter.
4801 * Fix bug in loading of registry value HKLM\SOFTWARE\OpenAFS\Client
4802 "EnableKFW". This value will not be read if the key
4803 HKCU\SOFTWARE\OpenAFS\Client exists; even if the "EnableKFW"
4804 value under that key does not.
4806 * provide mechanisms to force the use of krb524d for Kerberos 5
4807 ticket to AFS token conversion. For afslogon.dll and afscreds.exe
4808 there is a new registry value "Use524" and for aklog.exe a new
4809 command line parameter "-m".
4811 * Fix the pattern matching algorithm to properly match patterns
4814 * smb_ReceiveCoreRename() was factored to produce smb_Rename()
4815 which is used by both the original function and the new
4816 smb_ReceiveNTRename(). smb_ReceiveNTRename() supports the
4817 creation of HardLinks in addition to Renaming. smb_Link()
4818 is a new function which creates HardLinks via cm_Link().
4819 cm_Link() is a new vnodeops function which creates links
4822 smb_ReceiveNTRename() does not support the File Copy and
4823 Move Cluster Information operations described in its interface.
4824 ReceiveNTRename is under documented in CIFS-TR-1p00_FINAL.pdf.
4826 * When opening files via symlinks, we should follow the symlinks
4827 until we reach the actual file stat cache entry. The stat cache
4828 entry of the file should then be stored in the FID instead of
4829 stat scache entry of the symlink.
4831 * return bad operation errors for all unimplemented functions
4832 even if we do not know the functions exist.
4834 * Log bad packets and unknown operation packets to the trace log
4836 * Map CM_ERROR_BADOP to STATUS_NOT_SUPPORTED instead of
4839 * Update list of known CIFS operations to include all those listed
4840 in CIFS-TR-1p00_FINAL.pdf.
4842 * Modify the handling of HKLM\SOFTWARE\OpenAFS\Client\Submounts
4843 to support the REG_EXPAND_SZ type.
4846 * A new Windows authorization group "AFS Client Admins" is now
4847 created and populated with the members of the "Administrators"
4848 group. The group is used to determine which accounts on the
4849 machine may be used to modify the AFS Client Configuration via
4850 the UI and command line tools. afs_config.exe, fs.exe,
4852 * Modify the WinLogon Logoff Event Handler to query NT4 domain
4853 controllers for the remote profile path if Active Directory
4854 services are not available.
4856 * Fix aklog.exe to not add the AFS ID to the username
4858 * PTS registration of new users to foreign cells has been added to
4861 * The cm_Daemon thread is used to perform checks for
4862 down servers, up servers, volumes, callback expirations,
4863 lock maintenance and token expiration. Due to a gaff in
4864 larger integer division the thread never performed any
4865 work. Instead the current time computation would always
4866 be less then the trigger times. This had an adverse affect
4867 on the client's ability to maintain communication with servers,
4868 keep volumes up to date, and flush user tokens and acls
4869 when they have expired. This was broken when the 1.3 branch
4870 was modified to support VC7 which no longer included
4873 * An initialization problem with the Freelance code was
4874 detected while fixing the callbackRequest. The cm_rootSCachep
4875 object is obtained during afsd_InitDaemons() but the callback
4876 information is incomplete. The callback information will not
4877 be obtained until cm_MergeStatus is called from within
4878 cm_GetCallback. Unfortunately, cm_SyncOp did not properly
4879 test for the conditions under which the callback information
4882 * Reports have been filed indicating that callbacks were
4883 being lost. An examination of the code indicated that the
4884 cm_server_t objects were not being properly reference
4885 counted by the cm_scache_t and cm_callbackRequest_t objects.
4886 In particular, the cm_server_t objects may have been freed
4887 from beneath the cm_conn_t objects.
4889 All of the reference counting is now done via the functions:
4894 this improves the ability to track the referrals.
4896 Each cm_BeginCallbackGranting Call now allocates a reference
4897 to the cm_server_t. The cm_EndCallbackGrantingCall either
4898 frees the reference or transfers it to the cm_scache_t
4899 cbServerp field. These are then appropriately tracked
4900 through the cm_Analyze call.
4902 * Ensure that the dnlc hash table is the same size as the
4903 dir name hash table (as per original author's note).
4904 Increase the dnlc CM_AFSNCNAMESIZE to a multiple of 8
4905 for compatibility with 64-bit systems.
4907 * fix smb_ApplyV3DirListPatches to properly apply the hidden
4908 attribute to dotfiles when the infoLevel < 0x101 and
4909 cm_SyncOp has failed.
4911 * Fix the Freelance registry initialization code. There
4912 was a possibility that some systems could end up with
4913 garbage in the registry during a clean install.
4916 * file and directory names beginning with "." will now be given the
4917 hidden attribute when the volume access is anonymous. this matches
4918 the behavior when the volume access is via an authenticated user.
4920 * Added a change monitor to the HKLM\SOFTWARE\OpenAFS\Client\Freelance
4921 key. When a change occurs mark the root.afs data as invalid and
4922 for it to be reloaded on the next access. This allows administrators
4923 to modify the mount point list without restarting the service.
4925 The freelance client used to provide a fake modification time for
4926 the root.afs volume data and its mount points of 7/09/2001 14:24 EDT.
4927 Added code to extract the last modification time of the Freelance
4928 registry key and use that instead. The time now represents the
4929 most recent mount point change.
4931 * PTS registration of new users to foreign cells has been added to
4934 * Additional Cache Control and Credential Manager options have been
4935 added to the WiX installer. See deployment guide for details.
4937 * The CachePath setting is now optionally a REG_EXPAND_SZ type
4939 * The WiX installer has been upgraded. Version 2.0.1927.1 is now
4942 * The loopback installation code may have had a problem updating the
4943 %ETC%\HOSTS file which could have resulted in a premature failure.
4944 Work around code has been added for the case where the file cannot
4947 * The default max chunksize was increased from 15 (32K) to 17 (128K)
4948 because Windows sends 64K blocks when using overlapped writes.
4950 * The default number of server threads was increased from 4 to 25 to
4951 better handle overlapped writes.
4953 * The "AfscredsShortcutParams" registry value was not being properly
4954 loaded by afscreds.exe. Therefore, the default value was always being
4955 used instead of the value set by the installer.
4957 * Windows XP provides downgrade attack detection to prevent an attacker
4958 from being able to force the use of NTLM simply by disrupting
4959 communication with the KDC. This attack cannot exist between the
4960 Windows CIFS client and the AFS Client Service. Therefore, when a
4961 downgrade has been detected the afs pioctl library will force the
4962 establishment of a new CIFS connection using NTLM.
4964 * A locking error was discovered surrounding all references to volume
4965 server lists within the cm_cell.c source file.
4967 * The logged into Windows username was incorrect on Terminal Server
4970 * A new registry value "NonPersistentCaching" was added to the service
4971 parameters key. When set to a non-zero value, the afs cache is stored
4972 in the Windows paging file. There are two limitations to choosing
4974 1. when persistent caching is implemented it won't work with
4975 this flag set since there will be nothing to persist.
4976 2. with this flag set the initial paging allocation cannot be
4977 changed while the service is running
4979 * An initialization bug was discovered in aklog.exe which affected users
4980 who have a domain name for their afs servers which could not be mapped
4984 * afs_config.exe now validates cell names against DNS in addition
4985 to the CellServDB file.
4987 * In order to allow the freelance client to connect to a volume with ID
4988 equal to 1 on the default cell we changed the fake root.afs volume ID
4989 once again. This time we choose 0xFFFFFFFF. In addition, we change
4990 the cell ID of the fake root.afs volume from 1 to 0xFFFFFFFF as well.
4991 It will now be impossible for a volume ID to match that of another
4992 cell unless the client is connected to 0xFFFFFFFD cells. That should
4993 be enough room for growth.
4995 * Fix "fs mkmount" command to work with UNC paths and when
4996 started from non-AFS drives. It is now possible to create a mount
4997 point in the freelance fake root.afs volume with the command
4999 fs mkmount \\AFS\all\<directory-name> <volume-name> <cellname>
5003 fs mkmount \\AFS\all\openafs.org root.cell openafs.org
5004 fs mkmount \\AFS\all\.openafs.org root.cell openafs.org -rw
5006 * The algorithm used to re-attempt access to the servers associated with
5007 a volume has been altered to properly address the case in which all
5008 servers have been marked down. The previous algorithm did not reset
5009 the server's down flags so the servers were never actually retried.
5010 This caused a problem with active volumes if the network connectivity
5011 was lost as could be the case with a network cable removal, wireless
5012 drop, or laptop hibernation. With the fix volume access is restored
5013 almost instantenously when network connectivity becomes available.
5015 * Support for SMB/CIFS browsing has been added to the AFS Client Service
5016 SMB server. It is now possible to use "NET VIEW \\AFS" to obtain a
5017 listing of AFS submounts and freelance mount points. Support for
5018 NETSHAREENUM, NETSHAREGETINFO, NETSERVERENUM2, NETSERVERGETINFO
5019 significantly enhances the behavior of AFS volumes within the Explorer
5020 Shell. For instance, "AFS" now shows up as server in the Explorer
5021 with each submount or freelance mount point visible as a share.
5022 The right click menu in each folder now works with full functionality
5023 on a consistent basis.
5025 * The network provider can be configured to have different behavior
5026 depending on the domain that the user logs into. These settings are
5027 only relevant when using integrated login. A domain refers to an
5028 Active Directory (AD) domain, a trusted Kerberos (non-AD) realm or the
5029 local machine (i.e. local account logins). The domain name that is
5030 used for selecting the domain would be the domain that is passed into
5031 the NPLogonNotify function of the network provider. (see registry.txt
5034 * Added a new registry value [HKCU\SOFTWARE\OpenAFS\Client]
5035 "Authentication Cell" which may be used to specify a default
5036 authentication cell for afscreds.exe which is different from
5037 the default cell for the AFS Client Service daemon.
5039 * Added a Logoff WinLogon Event Notification function to afslogon.dll.
5040 afslogon.dll moved to %WINDIR%\System32\.
5041 New registry entries added to register the dll for Winlogon events.
5043 The logoff event will now force a call to ktc_ForgetAllTokens()
5044 using the context of the user being logged off as long as the
5045 user's profile is not loaded from within AFS. If the profile
5046 was loaded from AFS we can't release the tokens since the Logoff
5047 event is triggered prior to the profile being written back to
5048 the its source location. This is now performed in an XP SP2
5051 * Windows XP SP2 Internet Connection Firewall interoperability
5054 * The %WINDIR%\afsdsbmt.ini contains four sections:
5055 Submounts, Drive Mappings, Active Maps and CSC Policies.
5056 The Submounts and CSC policies are now stored in the registry under
5057 [HKLM\SOFTWARE\OpenAFS\Client\Submounts]
5058 [HKLM\SOFTWARE\OpenAFS\Client\CSCPolicy]
5059 The Drive Mappings and Active Maps are stored in the registry under
5060 [HKCU\SOFTWARE\OpenAFS\Client\Mappings]
5061 [HKCU\SOFTWARE\OpenAFS\Client\Active Maps]
5063 There is no automatic migration of this data as it would be impossible
5064 to consistently migrate data to user profiles which may not be active
5065 when the machine is updated.
5067 * The %WINDIR%\afs_freelance.ini contains lists of mountpoints for the
5068 fake root.afs volume. For the same reasons as for the cellservdb file,
5069 this information should not be in %WINDIR%. This information is now
5070 kept under the registry key
5071 [HKLM\SOFTWARE\OpenAFS\Client\Freelance]
5073 The data from the afs_freelance.ini file will be automatically
5074 migrated to the registry on first execution of afsd_service.exe
5076 * Keeping the CellServDB file in the location %WINDIR%\afsdcell.ini is
5077 troublesome for several reasons. One, it is confusing for those who
5078 expect the file to be named "CellServDB" instead of "afsdcell.ini".
5079 Two, this file is not a Windows Profile formatted file. Three,
5080 applications should not be reading or writing to %WINDIR%. It causes
5081 problems for Windows Terminal Server.
5083 The new location of CellServDB will be the OpenAFS Client install
5084 directory which is by default C:\Program Files\OpenAFS\Client and can
5085 be determined by querying the registry for
5086 [HKLM\SOFTWARE\TransarcCorporation\AFS Client\CurrentVersion]PathName
5088 The existing afsdcell.ini will be migrated by the NSIS installer.
5089 The Wix installer must still be updated to do the same.
5091 * Change NSIS installer to use DNS by default; to remove Integrated Logon
5092 High Security mode; and to add Terminal Services compatibility registry
5093 entries to allow the OpenAFS tools to find the afsdcell.ini and other
5094 configuration files in %WINDIR%.
5096 * Add support for authenticated SMB connections. This will remove
5097 the need for high security mode in most situations. Both NTLM
5098 and Extended Security (GSS SPNEGO) modes are supported. Effectively,
5099 only NTLM can be used even though Kerberos is now supported. The
5100 reason is that it is not possible to construct a service principal
5101 which is unique to each individual machine.
5103 SMB Extended Auth does not work on XP SP2 unless one of two registry
5104 modifications are made:
5106 (1) To disable the check for matching host names on loopback connections
5107 set this key. This does not require a reboot:
5109 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa]
5110 "DisableLoopbackCheck"=dword:00000001
5112 (2) To add the AFS SMB/CIFS service name to an approved list. This
5113 does require a reboot:
5115 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa\MSV1_0]
5116 "BackConnectionHostNames"=multi-sz "AFS" "MACHINE-AFS"
5118 afsd_service.exe will automatically add the current Netbios Name
5119 to the BackConnectionHostNames list and then temporarily disable
5120 the loopback check for one cycle of startup/shutdown of the service.
5121 We assume most folks do not start/stop without a reboot so this
5122 will be adequate in most cases.
5124 * Fix security hole in afslogon.dll which allowed passwords to be
5125 sent in clear text to the KDC in a misformed principal name.
5127 * Fix cm_GetCell() to properly handle expired dns entries
5130 * If Freelance mode is active and the afs_freelance.ini
5131 file does not exist, do not create an empty file.
5132 Instead create a file containing ro and rw mountpoints
5133 to the default cell using the standard conventions.
5135 * Modify the Freelance support to handle the ability
5136 to create rw mount points in the fake root.afs volume.
5138 * Changed the RPC mechanism used for token setting from
5139 named pipes to local. Use of named pipes can be restored
5140 by setting the environment variable AFS_RPC_PROTSEQ to
5143 Named pipes were required when a Windows 9x system was
5144 using a NT system in gateway mode which is incompatible
5145 with our use of local loopback adapters.
5147 * In afscreds.exe, if a username of the form user@REALM is
5148 specified and no password is specified, do not perform a
5149 kinit operation. Only perform the aklog functionality.
5151 * Add a new registry value which allows the number of processors
5152 on which afsd_service.exe executes to be restricted. Valid
5153 values are 1..numOfProcessors
5155 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
5159 * A second MSI based installer option is now available.
5161 * Fixed Kerberos 5 kinit functionality in afscreds.exe to properly
5162 request tickets for user/instance@REALM instead of just user@REALM
5164 * Modify the Power Management Notify routine to wait for the Hard Dead
5165 timeout period instead of a fixed 19 seconds. With the longer timeout
5166 periods Hibernation and Standby could never succeed when network
5167 connectivity is not available.
5169 * The following fs.exe commands are now restricted to Administrator:
5170 - checkservers with a non-zero timer value
5173 - sysname with a new sysname list
5182 setting the default sysname for a machine should be done via the
5183 registry and not via "fs sysname".
5185 * NSIS installer adds options to install Debugging Symbols
5186 and the Microsoft Loopback Adapter; the user is now also
5187 given the ability to select the afscreds.exe startup options.
5189 * Build system modified to generate symbols for FREE (aka RELEASE)
5190 builds as well as CHECKED (aka DEBUG) builds
5192 * Sites which have a volume ID of 0x20000001 assigned to their
5193 root.cell volumes have been experiencing problems with accessing
5194 the root.cell volume of their cell when Freelance mode has been
5195 active. This was because 0x20000001 was assigned to the fake
5196 root.afs volume created by freelance. The fake volume id is
5197 now set to 0x00000001 to prevent conflicts.
5199 * The timeout logic in the AFS Client Service has been wrong
5200 for sometime. It is based on two different assumptions.
5201 First, the SMB client timeout is a fix value as was the case
5202 with OS/2 Lan Manager. This assumption is incorrect. The
5203 SMB timeout in Windows is a dynamic value computed based upon
5204 a fixed minimum timeout to which is added time based upon the
5205 size of the request and the performance characteristics of
5206 the connection. Second, it is the responsibility of the
5207 SMB Server to enforce the timeout requirements of the client.
5208 This is untrue. The SMB Server cannot be expected to know
5209 the requirements of the client. More importantly, if the
5210 SMB server uses the SMB client timeout as a value to restrict
5211 its behavior as an RX client, the performance characteristics
5212 of the local SMB session would be used to prematurely terminate
5213 WAN connections with significantly different performance
5216 The timeout logic has therefore been modified in the following
5218 . the Lan Manager Workstation (SMB) Session Timeout is used only
5219 as a basis for configuring the Connection Dead Timeout
5220 and Hard Dead Timeout values. The Connection Dead Timeout
5221 must be at least 15 seconds longer than the SMB Timeout
5222 and the Hard Dead Timeout must be at least double the
5223 Connection Dead Timeout.
5224 . New registry entries have been added to allow the Connection
5225 Dead Timeout and Hard Dead Timeout values independent of the
5226 Lan Manager Workstation Session Timeout
5227 . The test to enforce the SMB Client Timeout has been removed.
5229 One of the side-effects of removing the enforcement of the SMB
5230 Client Timeout is that regardless of whether or not the SMB client
5231 is available to receive the response (and how would the SMB server
5232 know) the RX protocol response can be used to update the AFS
5233 Client Service state for ready access by future SMB client
5236 This should be the end of the "Server paused or restarting messages"
5238 * Add support for arbitrary UNC paths to the pioctl() support.
5239 This enables the fs commands as well as the AFS Shell Extension
5240 to work correctly when UNC paths are being used.
5242 * Fix afscreds.exe (by updating afskfw.lib) to search for cells via
5243 DNS if the cell configuration cannot be determined via CellServDB
5245 * Add debug info to test whether CM_BUF_WAITING or CM_SCACHE_WAITING
5246 are ever set more than once at a time
5248 * Fix the management of lists of cm_cell_t structures when using
5249 DNS to lookup cell information. The previous code would fail to
5250 reuse the same cellID for a cell if DNS was used more than once
5251 for a given cell name. When the ttl expired, a single cm_cell_t
5252 could be inserted into the cm_allCellsp list more than once
5253 producing a loop. In addition, the vlServerp list belonging to
5254 the cell was not freed resulting in improper refCounting of the
5257 * Add DNS support to cm_IoctlNewCell() which previous only examined
5260 * Add cm_FreeServer() function and call it from cm_FreeServerList()
5261 to properly garbage collect cm_server_t objects
5263 * Add numVCs variable to smb.c to track the number of smb_vc_t
5264 objects created and use it to initialize the vcID field which
5265 previously was set to 0 in all objects resulting in FindByID
5268 * Fixed DNS lookups to work consistently throughout the OpenAFS
5269 product instead of just from within the afsd_service.exe
5271 * Added a runtime check to ensure that AFS Client Service SMB
5272 Server is accepting connections before attempting to mount
5275 * Read IP addresses for volume servers out of the CellServDB
5276 file if gethostbyname() on the hostname fails.
5278 * Fix getcellconfig() to populate both the Hostnames as well
5279 as IP addresses when loading cell data via DNS
5281 * Increase the Connection Dead Time to 50 from 20 seconds
5282 Increase the Hard Dead Time to 120 from 40 seconds
5283 (matches the Unix values)
5285 * Fixed an assertion validating the number of allocated NCBs
5287 * Fixed the build environment to consistently build for
5288 Windows 2000 and above. (APPVER = 5.0)
5290 * Fixed rx_debug to properly validate the receipt in incoming
5291 data with select() and recvfrom(). Do not copy data out of
5292 the socket buffer unless success is indicated.
5295 * afsd_service.exe will now display a message box to the
5296 desktop when it terminates due to an IP Address Change.
5298 * installer no longer deletes AFS Server configuration data
5301 * installer generates a warning dialog if the RPC service
5302 is not properly configured
5304 * installer compressed with lzma instead of bzip2
5306 * afsd_service.exe shutdown crash solved once and for all
5308 * reference counting of smb_vc_t data structures improved
5310 * name space collision of smb_fid_t event objects corrected
5312 * the output of "fs memdump" is now written to
5313 %WINDIR%\TEMP\afsd_alloc.log
5315 * the file TaAfsApp_1033.dll is now properly installed allowing
5316 the User Manager to start
5318 * a new algorithm is used for computing filename pattern matches
5320 * afscreds.exe now accepts user names containing instance
5323 * Fix the Directory Name Lookup Cache to be case-sensitive.
5324 This is crucial in environments in which a Windows client
5325 is accessing a directory with more than one filename that
5326 differs only by case. If the directory contains "FOO"
5327 and "Foo". You want "DEL Foo" to delete the correct one.
5328 We still have a problem in that "DEL foo" will delete a
5329 random filename. This will be addressed in a future release.
5331 * Fix afscreds.exe -M option (renewMaps) to work when High
5332 Security mode is off. Also, remember to disable the ActiveMap
5333 flag in afsdsbmt.ini when a drive mapping is removed.
5335 * Updates to NSIS installer script. AFS Server configuration
5336 data will not be destroyed on un-install or re-install.
5337 Use a better compression algorithm.
5339 * afslogon.dll now uses KFW to obtain tokens when available
5341 * afslogon.dll when given an all uppercase username will
5342 attempt to authenticate with both the uppercase name
5343 and an all lowercase variation
5345 * DST modification removed. The fix appears to make things
5346 worse after a reboot of the machine.
5348 * fs.exe: added "cscpolicy" which is used to
5349 change client side caching policy for AFS shares
5351 Usage: fs cscpolicy [-share <AFS share>] [-manual] [-programs]
5352 [-documents] [-disable] [-help]
5354 * Several uninitialized variables have been initialized
5356 * It is now possible to obtain tokens using cross realm
5357 Kerberos within afscreds.exe:
5359 user: jaltman@ATHENA.MIT.EDU
5361 Will obtain a cross realm ticket for jaltman/DEMENTIA.ORG@ATHENA.MIT.EDU
5362 will will in turn be used to obtain afs@DEMENTIA.ORG.
5363 The resulting token will be stored with the display name
5364 jaltman@ATHENA.MIT.EDU@dementia.org
5366 * aklog.exe has been added to the client
5368 Usage: aklog [-d] [[-cell | -c] cell [-k krb_realm]]
5369 [[-p | -path] pathname]
5373 -d gives debugging information.
5374 krb_realm is the kerberos realm of a cell.
5375 pathname is the name of a directory to which you wish to authenticate.
5376 -noprdb means don't try to determine AFS ID.
5377 -5 or -4 selects whether to use Kerberos V or Kerberos IV.
5378 (default is Kerberos V)
5379 No commandline arguments means authenticate to the local cell.
5382 * All of the resource files have been restructured to adhere to
5383 a set of rules IBM implemented for loading string resources.
5384 These rules had either been forgotten or were not discovered
5385 by folks working on the OpenAFS sources. The end result was
5386 memory corruption. This is primary item which was preventing
5387 the AFS Server from working.
5389 * Increased the size of the maximum ticket size stored in a token
5390 from 344 bytes to 12,000. Increased the buffers used to convey
5391 messages between the pioctl() caller and the SMB Server from
5392 1000 bytes to 12,512. The code appeared to have been writing
5393 above the top of the stack by quite a few number of bytes.
5394 (The increased ticket size is necessary for the next item.)
5396 * When obtaining AFS Tokens via KFW, krb524 is no longer required.
5397 Instead the raw Kerberos 5 ticket is used in its entirety. This
5398 is extremely important as it allows us to use pure Kerberos 5 KDCs
5399 as the source of the AFS authentication. The use of up to 12,000 byte
5400 tickets will allow tickets produced by all versions of Microsoft
5401 Active Directory to be used.
5402 - create a user account.
5403 - designate it DES only
5405 - specify its UPN to be "afs@realm"
5406 - assign a SPN of "afs/cellname" to the UPN with setspn.exe
5408 * Do not enforce the funky 8dot3 pattern matching rule that the first "."
5409 is special when using long file names. (you must use "*.*" and not "*")
5410 Instead only enforce it when performing 8dot3 searches.
5412 * Fixed the DST problem with creation times being set one hour ahead
5414 * Fixed the problem when using \\afs\cell-alias. For example,
5415 \\afs\uncc instead of \\afs\uncc.edu. Do not a new cell struct
5416 for the alias name; instead simply expand the name. One of the
5417 symptoms of this problem was a loss of acquired tokens.
5419 * Fixed the AFS Shell Extension. The Symbolic Link menu was empty
5420 of strings. (Only English strings provided.)
5422 * Fixed the installer to properly replace in use files.
5424 * Fixed the build system to cleanup generated component version files
5426 * The release build compiled with MSVC 6.0 compiler to avoid the
5427 afsd_service.exe shutdown crash. This does not solve the problem
5428 but simply avoids it for the time being.
5432 * fix afslogon.dll to not corrupt memory when High Security mode
5435 * fix afsd_service.exe to not attempt to restore the stack when
5436 an exception occurs. (not safe in multi-threaded programs)
5438 * fix uninstaller to properly remove the CRT and MFC DLLs
5440 * remove a Message Box from afscreds.exe when getcellconfig()
5441 fails on a kerberos realm which is not a cell
5443 The following is a list of changes to the OpenAFS for Window client
5446 * "fs setserverprefs" will leave afsd service deadlocked
5448 * "vos listaddrs" will core dump
5450 * installer sets the appropriate keys to support Integrated Logon
5452 * installer disables the "Find Lana by Name" functionality as it
5453 was causing headaches for many users
5455 * fix the intermittent crash of the power management thread when
5456 shutting down the AFS Client Service
5458 * optimizes the obtain drive mount list functionality which is
5459 executed every time the mount tab in afscreds.exe and afs_config.exe
5460 are refreshed. (this happens a lot)
5462 * fix the service shutdown logic. add the STOP_PENDING state
5463 and do not accept additional service events after we declare
5466 The following is a list of changes to the OpenAFS for Window client
5469 * flexelint was run against the source tree and hundreds (perhaps
5470 thousands) of corrections were applied to ensure prototypes
5471 were in use; types were used consistently; variables were
5472 initialized; unused variables were removed; etc.
5474 * A wide variety of instrumentation was added including the
5475 ability to produce a stack trace from within afsd_service.exe
5478 * Dynamic configuration of the RDRtimeout value based upon the
5479 LanMan Workstation Session Timeout
5481 * The mount root no longer needs to be called "/afs". This
5482 is now set by a registry value "MountRoot" within the key
5483 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5485 * The cell list is now only read out of afsdcell.ini when the
5486 file changes instead of each time a cell is resolved.
5488 * Thread synchronization was added to cm_server.c and ktc_nt.c
5490 * All calls to GlobalAlloc()/GlobalFree() were replaced with
5491 calloc()/free(). The Global functions were needed on Windows 3.x
5492 but have caused a variety of problems on the Win32 platforms.
5493 Avoiding them is highly recommended by several Microsoft
5494 Knowledgebase articles
5496 * Support for Symbolic Links added to the AFS Shell Extension
5498 * Added a registry value "OverlayEnabled" to determine if
5499 Shell Extension Overlays should be enabled.
5500 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5502 * New Build system to support VC6, VC.NET, VC.NET2003 compilers and
5503 separate trees for checked and free builds. Build system supports
5504 a custom directory src\WINNT\extra which can be used as a grafting
5505 location of organization specific additions to the build tree.
5507 * New installer built using NSIS 2.0.
5509 * Named all kernel objects in order to allow them to be monitored
5510 with tools such as SysInternals' ProcExp.exe.
5512 * Introduced new EventLog framework for AFSD
5514 * Introduced Power Management interface to AFSD for Standby and
5515 Hibernate modes to allow cache to be flushed prior to network
5518 * Utilize Win32 DNSQuery API instead of internal routines. This
5519 allows DNS SRV queries to be sent to all current domain name
5520 servers. Not just one specified in an INI file. DNS is now
5523 * "NetbiosName" registry value may be used to specify a fixed
5524 Netbios Name such as "AFS" to be used instead of "HOSTNAME-AFS"
5525 when the loopback adapter is in use. If you need to use the
5526 old notation with a loopback adapter installed specify a registry
5529 "NetbiosName" REG_EXPAND_SZ = "%COMPUTERNAME%-AFS"
5531 * Refactor all modules which depend on LAN Adapter and NetbiosName
5532 determination in a new library: lanahelper.lib. This allows for
5533 consistent behavior throughout the product.
5535 * Move the afsd.log and afsd_init.log files to the directory specified
5536 by the "TEMP" environment variable. This is usually %WINDIR%\TEMP
5537 for services. Added the Date to the log entries.
5539 * New registry value "RxMaxMTU" used to limit the size of the RX
5540 packets sent by the AFS Client Service to the Server. In order
5541 to enable OpenAFS to work across the Cisco IPSec VPN the packet
5542 size must be restricted to 1264 or smaller. The latest NSIS
5543 installer sets a value of 1260 by default.
5545 * New registry value "RxNoJumbo" to disable the use of Jumbo Rx
5546 packets. This is not needed in order to work across the Cisco
5547 VPN but might be needed for other network environments. This
5548 value is not set by the NSIS installer.
5550 * New registry value "HideDotFiles" is used to apply the Hidden
5551 attribute to files whose names begin with a '.'. This value
5552 is set by the NSIS installer.
5554 * New registry value "MaxMpxRequests" allows the maximum number
5555 of multiplexed sessions to be configured at run time. This
5556 value is not set by the NSIS installer. The default value is
5559 * New registry value "MaxVCPerServer" allows the maxmimum number
5560 of VCs per server to be configured at run time. This value is
5561 not set by the NSIS installer. The default value is 100.
5563 * New registry value "AllSubmount" allows the "all" submount to
5564 be disabled by setting its value to 0x00.
5566 * Allow cells names to be valid mount points
5567 \\<netbiosName>\<cellname>
5569 * Store the active state of drive mappings in order for afscreds.exe
5570 to restore them upon startup
5572 * Add exception handling to generate a Stack Trace to the afsd_init.log
5573 file if one happens to occur.
5575 * Add lots of logging to help detect the cause of invalid SMB packets
5577 * Enable Kerberos for Windows to be used to obtain AFS Tokens via
5578 conversion of Kerberos 5 "afs" service tickets. Supports auto-
5579 renewal of expiring tokens as long as afscreds.exe is running.
5581 * New afscreds.exe command line options:
5583 -M = renew drive maps
5584 -N = ip address change detection
5587 * New registry value "EnableKFW" in {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5588 determines whether or not MIT Kerberos for Windows should be used
5589 to obtain tokens via Kerberos 5 tickets.
5591 * New registry value "AfscredsShortcutParams" in
5592 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5593 determines the command line parameters to be specified when "fixing"
5594 the AFS Shortcut in the user's startup folder.
5596 * The "ShowTrayIcon" registry value has been moved from
5597 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5598 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5600 * The <cell name> registry values used to store the token expiration
5601 reminders have been moved from
5602 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5603 {HKCU,HKLM}SOFTWARE\OpenAFS\Client\Reminders
5605 * Obtain the Logon User Name from the Explorer key when available
5607 * new text document doc\txt\winnotes\registry.txt lists all registry
5608 values used by OpenAFS (excluding the AFS Server)
5610 * BUG: rx_securityClass objects were not properly reference
5611 counted and were never freed.
5613 * BUG: reduce the number of conditions under which CM_ERROR_TIMEOUT
5614 would be generated. The existence of a server does not imply
5615 that it is not down. If all of the servers for a cell are down
5616 return CM_ERROR_NOSUCHVOLUME instead. This prevents the Explorer
5619 * BUG: the directory name lookup cache failed to free the entries
5620 in the cache when the name cache entries cycled. The entries
5621 in the cache would become dereferenced without being freed.
5623 * BUG: fs setserverprefs could be executed without Administrator
5626 * BUG: the number of allocated NCB objects (100) exceeded the number
5627 which could actually be waited upon by the kernel (64). Any objects
5628 which were utilized above the limit could never have event completions
5631 * BUG: smb_username_t objects were not being reference counted and
5632 were not properly freed.
5634 * BUG: smb_tid_t objects could under unusual circumstances be freed
5635 before they were no longer referenced.
5637 * BUG: smb_fid_t object pointer were frequently used even when
5638 their value could be NULL. They were not properly released and
5639 therefore they were never freed.
5641 * BUG: smb_packet_t data structures were not completely initialized
5644 * BUG: when Rx produces a CM_ERROR_NOIPC error do not return "Access
5645 Denied" because that causes the Explorer Shell to try again until
5646 access is obtained. Instead return "Remote Resources" which allows
5647 the shell to move on and treat the error as transient.
5649 * BUG: when initializing the NCBreturns structure, separate Event objects
5650 were created for each NCB although a single Event object was supposed
5651 to be shared by all.
5653 * BUG: smb_dirSearch_t objects were not being properly referenced counted
5656 * BUG: smb_tran2Packet_t objects were not being properly referenced
5659 * BUG: directory path creation did not handle the case of multiple
5660 directories requiring creation in one attempt
5662 * BUG: SMB requests which required an Extended Response were ignored.
5663 This prevented some files from being written to AFS volumes.
5665 * BUG: character strings were being freed even after they were
5666 inserted into in use data structures
5668 * BUG: inconsistent usernames were used when High Security mode was
5669 enabled. (there is still much to do in this area)
5671 * BUG: pioctl() calls which require out of band RPC operations were
5672 susceptible to race conditions when performed by multiple processes
5674 * BUG: memory allocation and deallocation crossed instances of the
5675 C Runtime Library producing memory leakage and corruption in
5676 afscreds and the client configurator.