2 * If a file server becomes inaccessible while the cache manager has
3 dirty buffers to write, the afsd_service buf_IncrSync thread can
4 attempt to use 100% of the cpu.
6 * Fix the locking that protects the cm_cell_t lists
9 * Remove cm_cell_t objects from the name and id hash tables before
10 placing them on the free list.
12 * Fix "fs newcell" which was broken in 1.5.60.
14 * Do not attempt to synchronize dirty buffers if the associated
15 volume is known to be unavailable.
17 * Log to the Windows Event Log dirty buffers that cannot be written
18 during the shutdown process.
20 * Modify behavior of a Freelance mountpoint target that does not
21 specify a cell. Instead of assuming the target volume is in the
22 Freelance.Local cell, use the workstation "Cell" specified in the
23 registry. A mountpoint target of "#root.cell." will now mean the
24 root.cell volume in the workstation cell for the current session.
25 If the workstation cell changes from "athena.mit.edu" to
26 "andrew.cmu.edu", the referenced volume will also change without
27 requiring that the mount point targets be altered.
29 * Avoid false infinite loop errors when validating the cm_cell_t
30 free list during startup.
32 * Fix two more locations where xdr_free() should be used instead of
33 free() to permit a checked built of afsd_service.exe to work with
34 release builds of afsrpc.dll.
36 * Add cm_FindServerByUuid(). Re-implement RXAFS_InitCallBackState3()
37 to permit the server Uuid to be used to lookup the server object
38 and from that determine the cell. This permits callbacks that are
39 received from alternate addresses to be processed with a known server
40 object. Previously a request from an unknown server would clear all
41 callbacks from all cells.
43 * Fix a bug that prevented optimal performance when using a non-zero
44 value for 'daemonCheckVolCBInterval'. As a reminder, when
45 "daemonCheckVolCBInterval" is set to a non-zero value, all .readonly
46 volume callbacks are automatically renewed 90 minutes before their
49 * Fix automatic ranking of vldb servers whose values are obtained from
52 * Add logging for RX CALL DEAD errors which are generated if the remote
55 * Add failover for RX CALL TIMEOUT errors when the volume is readonly
56 or the call is to a vldb server.
58 * Add registry based cell search functionality to NetIdMgr,
59 afs_config.exe, and klog.exe.
61 * afsconf_GetCellInfo() has been modified to perform gethostbyname()
62 lookups on the host names in the CellServDB instead of using the
63 specified IP addresses. This provides aklog, pts, vos, etc. the same
64 CellServDB behavior that the Windows Cache Manager uses.
66 * Completely a behind the scenes change. The snprintf() routines have
67 been completely rewritten.
69 * When updating the stat cache entry callback of a .readonly object
70 from the volume group object, update the file server reference to
71 ensure it matches the most update to date callback.
73 * Add proper support for processing callbacks from multi-homed file
74 servers. Instead of comparing servers by cm_server_t pointer,
75 compare them by UUID when the UUID is known.
77 * During a shutdown short circuit the offline volume check daemon
80 * Properly track link counts during cm_Link and cm_Unlink operations.
82 * Return the error code of RXAFS_FetchData / RXAFS_StoreData in
83 preference to an error code reported by rx_EndCall.
85 * Add "PerFileAccessCheck" registry value to permit testing against
86 experimental file servers that include per-file acl support. This
87 value is intentionally undocumented. It is not to be used by
88 production environment deployments.
90 * Log a warning to trace log if a volume group is marked as a DFS
91 File Set. The Windows OpenAFS client does not include DFS support.
93 * Fix a bug introduced in 1.5.60 that prevents the afs netidmgr
94 provider from obtaining tokens when referrals are in play.
96 * Add "fs chown" and "fs chgrp" commands to permit the owner and
97 group of objects stored in AFS to be set from Windows.
99 * Avoid performing background daemon operations when the machine is
100 going into suspend mode.
102 * Perform offline volume checks in most recently used order.
104 * Prevent crash when a data version for a cache object goes backwards.
106 * fs setquota, fs setcachesize, vos setfields, and vos create now
107 accept human readable orders of magnitude. (K, M, G)
109 * fs listquota fixed to permit large quota sizes to be displayed.
111 * rx packet resend and data packets sent counts were incorrect.
113 * Multi-thread safe library versions are now being generated and used.
114 mtafsubik.lib, mtafsutil.lib, mtafsvldb.lib, mtafsvol.lib.
116 * Microsoft SMB Redirector (mrxsmb.sys) support for ExtendedSessTimeout
117 values are now available on XP through Windows 7. Add functionality
118 to autodetect if such support is present on the machine. If so,
119 configure it if necessary and dynamically adjust the AFS Rx timeout
122 * Correct another race condition in the Rx library that could result
123 in an unexpected panic while freeing the Rx call iovq.
126 * A fix to the pioctl library to support drive substitution
127 to UNC paths. (SUBST <d:> <\\afs\cell\path>).
129 * On April 9th Microsoft released a Hot Fix for Windows Server 2003 SP2
130 that corrects a deadlock in the smb redirector and also adds new
131 functionality that permits the AFS SMB server to be given a longer
132 timeout than is normally the case. New functionality has been added
133 to configure these additional LanmanWorkstation\Parameter values.
134 (This functionality has been backported to XP SP3 and is scheduled
135 to be released on June 5th.)
137 * The BackConnectionHostNames registry value configuration was broken
138 when dynamic re-establishment of Netbios Name registrations was added.
141 * Hidden vos commands are revealed.
143 * If the "DisableLoopbackCheck" registry value is set, do not unset
144 it during the same service session.
146 * Reorganize code that prevents multiple Store operations to the same
149 * Modify IsPathInAfs test in Explorer Shell Extension and fs.exe to
150 permit broken symlinks to be treated as being in AFS.
152 * Fix vos commands that output 64-bit integer values.
154 * Cygwin Import Libraries are provided in the SDK for all OpenAFS DLLs.
155 Permits building cygwin applications against OpenAFS libraries.
157 * OpenAFS Release Notes, Administrator Guide and User Guide now installed
158 as Windows HTML Help (.CHM) files.
160 * NSIS installer does a much better job of cleaning up files left
161 over from previous installs.
163 * Fix RT#124787, a race condition between "fs flush <dir>", "fs flushvolume",
164 or "fs flushall" and on-going directory operations that can result in
165 afsd_service.exe crashing.
167 * Add support for DNS SRV records in place of AFSDB records.
168 _afs3-vlserver._udp.<cellname>. Priority field is used.
169 Weight is currently ignored.
171 * Add a method of specifying Client CellServDB information within the
172 registry that can be used to either override the CellServDB file or
173 force the use of DNS lookups for a given cell.
175 The registry schema is as follows:
177 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\
178 "LinkedCell" REG_SZ "[cellname]"
179 "Description" REG_SZ "[comment]"
180 "ForceDNS" DWORD {0,1}
182 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\[servername]\
183 "HostName" REG_SZ "[hostname]" Default: [servername]
184 "IPv4Address" REG_SZ "[address]" Used only if gethostbyname() fails.
185 "IPv6Address" REG_SZ "[address]" <future>
186 "Comment" REG_SZ "[comment]"
187 "Rank" DWORD "0..65535" Default: 0
188 "Clone" DWORD "{0,1}" <future: server only>
189 "vlserver" DWORD "7003" <future>
190 "ptserver" DWORD ... <future>
192 ForceDNS is implied non-zero if there are no [servername]
193 keys under the [cellname] key. Otherwise, ForceDNS is zero.
194 If [servername] keys are specified and none of them evaluate
195 to a valid server configuration, the return code is success.
196 This prevents failover to the CellServDB file or DNS.
198 Only one of "HostName" or "IPv4Address" is required. "HostName"
199 is optional if [servername] is the host name.
201 * Extend registry based CellServDB functionality to
202 afsconf_GetCellInfo() interface used by aklog and many
203 other command line utilities.
205 * libafsconf.dll moved from OpenAFS\Client\Program to OpenAFS\Common
206 as it is now used by both client and server components.
209 * PriorityClass of afsd_service.exe process raised to
210 "High" to match the priority of the system services
211 that are waiting for afsd_service.exe to respond.
212 Configurable using [TransarcAFSDaemon\Parameters]
213 DWORD "PriorityClass" value. (use values from Win32
216 * Change error reporting of CM_ERROR_BADSHARENAME to
217 use the correct error (NTSTATUS_BAD_NETWORK_PATH).
218 The error we were sending would cause the smb
219 redirector to drop the connection.
221 * Change the error reporting of CM_ERROR_ALLOFFLINE
222 and CM_ERROR_ALLDOWN to return NTSTATUS_PATH_NOT_FOUND.
223 Previous we would return Bad Network Path which
224 would cause the smb redirector to cancel all file
225 handles on the same "share".
227 * A debug version of afsd_service.exe can now be used
228 with a release version of afsrpc.dll (or vice versa).
229 afsrpc.dll now implements and exports xdr_free().
231 * Removed all unprotected OutputDebugString() calls as
232 those calls can block for up to 10 seconds if there
233 is contention for the globally shared memory or if
234 the debugger fails to read data in a timely manner.
236 * Visual Studio 2008, WDK 6.1 and DDK 6.1 are now
237 supported for builds. The production releases still
241 * Discovered that XP/2003 supports a variant on recvmsg
242 called WSARecvMsg. The input parameters are mostly
243 compatible with the Posix recvmsg. This will save
244 multiple memcpy() calls per received packet.
246 * Discovered that Vista/2008 supports a variant on
247 sendmsg called WSASendMsg. The input parameters
248 are mostly compatible with the Posix sendmsg. This
249 will save multiple memcpy() calls per sent packet.
251 * Exporting rx_ReadProc32, rx_WriteProc32 from afsrpc.dll
252 and utilizing rx_ReadProc32 from within the cache manager
253 as part of the FetchData operations. rx_ReadProc32 and
254 rx_WriteProc32 are optimized versions of rx_ReadProc and
255 rx_WriteProc for use when reading 32-bit integer values.
257 * The Network Identity Manager AFS provider now reports
258 progress events. These provide a better user experience
259 with Network Identity Manager 2.0.
261 * [RT 124293] A race condition exists which permits the
262 scp field of the an smb_fid_t object to become invalid
263 after a request on the smb_fid_t has begun. Remove the
264 race by protecting all access to the scp field with the
265 smb_fid_t mx mutex and obtaining a local reference on
266 the cm_scache_t object for the length of the request.
268 * [RT 124276] If the vldb is out of sync with the contents
269 of the file servers, afsd_service will retry too many
270 times when a file server reports a volume as not being
271 present. Now if the list reported by the vldb is the
272 same as the previously seen list, then the retry is
275 * [RT 124276] Read-only volume failover was broken in
276 1.5.53 whenever accessing a volume results in VNOVOL or
279 * [RT 124276] Prior to 1.3.70 the volume server reference
280 list was not reference counted and would be prematurely
281 freed while in use. When reference counting was added
282 in 1.3.70 a bug was introduced that could result in
283 service reference list corruption.
285 * Add Windows Application Event Log warning messages for
286 "Client SMB MPX value too large" and
287 "Client SMB Buffer Size too small".
289 * Renaming of files across directory boundaries would result
290 in an invalid handle error when attempting to access the
291 files after the move.
293 * Fix the handling of Tran2 Set Path Info RPCs. Do not fail
294 when a smb file descriptor cannot be found. The whole point
295 of using a Path Info function is because an smb file
296 descriptor wasn't allocated.
298 * More edge cases in which dynamic addition of Freelance
299 root.afs entries would get the wrong FID or where the
300 root.afs directory would not be refreshed.
302 * Buffer overflow could occur if the workstation cell name was
303 longer than 64 characters. Crashes could occur in
304 afscreds.exe, afslogon.dll, and afsd_service.exe.
306 * VNOSERVICE and VOFFLINE errors were leaking and were exposed
309 * Improve consistency of the bad data version value used for
312 * Log file server uuid values as part of the cm_server object
313 when available. Dump the cm_server object list in response
316 * Optimize the performance of resetting access control lists
317 when tokens are set or removed.
319 * Fix vos.exe so that it can work with very large volume
322 * Remove symlink recursion tests and increase max symlink count
326 * Undo the "UAC manifest fix" applied to afs_config.exe.
328 * Fixed some additional race conditions in the Rx library
330 * Ensure that Freelance allocation of vnodes follow
331 the AFS convention of odd vnodes are directories
332 and everything else is an even vnode.
334 * Add Freelance logic to mount point and symlink
335 evaluation functions.
337 * Enhance smb_ParseASCIIBlock() so that it can handle
338 all of the STRING formats defined by the CIFS Technical
341 * Validate the output of smb_ParseASCIIBlock() in all
342 callers. Return CM_ERROR_BADSMB if the STRING field
343 cannot be parsed. CM_ERROR_BADSMB will cause the
344 contents of the packet to be logged.
346 * If multiple SMB Raw Write operations were taking
347 place at the same time, there could be data corruption
348 because unique event objects were not generated for
349 each Netbios receive operation.
353 * Fix a race in rxi_WriteProc which could result
354 in a freed or in use packet being passed to
355 rxi_PrepareSendPacket if the call is currently
356 in a transmit wait state. Passing a freed packet
357 will result in a panic. Passing an in use packet
358 might result in a call failure or perhaps data
361 * Kerberos referrals broke the NIM AFS provider
362 logic for automated configuration of the workstation
363 cell with a new identity. Now if a referrals
364 response is returned, assume the cell belongs to the
367 * The interface to buf_CleanAsync() and
368 buf_CleanAsyncLocked() has been changed to permit
369 errors to be propagated to the caller.
371 * During an enumeration of a directory, if the btree
372 has not yet been generated, return an error.
374 * When mapping CM_ERROR values to NTSTATUS values
375 do not treat 0 as an error.
377 * Avoid a cm_scache_t object reference count underflow
378 when renaming via the SMB interface.
380 * smb_IsMask() can be called with a non-8.3 input buffer.
381 Separate smb_IsMask() into smb_Is8Dot3Mask() for 8.3
382 input buffers and smb_IsMask() for non-8.3 input buffers.
384 * Add cm_buf_t error state to the "fs memdump" output.
386 * pioctl drive mapping checks need to take into account
387 Global Drive Mappings and SUBST assignments.
389 * Rename operations should not be permitted if the
390 destination name collides with an existing name that
391 differs only by case UNLESS the collision is the object
392 that is being renamed.
394 * Increase the maximum number of background daemon threads
395 to 64. Leave the default at 4.
397 * Set the FILE_UNICODE_ON_DISK flag when reporting the
398 file system characteristics.
400 * Report the maximum component name length as 255 even
401 for Unicode as part of the file system characteristics.
403 * Add RX debugging information to "fs memdump" output
406 * Refcount the cm_server_t objects returned by
409 * Fix the UAC manifest applied to afs_config.exe
411 * rxdebug now indicates if any RX_CONN flag is set
412 not just some of them
414 * Change the mapping of CM_ERROR_WOULDBLOCK from
415 NTSTATUS_LOCK_NOT_GRANTED to NTSTATUS_CANT_WAIT
419 * Includes #GCO Public CellServDB 07 Nov 2008
421 * Prevent separate cell entries from being created for
422 "foobar.org" and "foobar.org." which can happen if
423 an AFSDB record is found instead of a CellServDB
426 * Fix a panic when checking the SMB virtual circuit list.
427 If a VC is destroyed while the list is being checked
428 it is possible for the next object to become invalid.
430 * Increased logging to afsd_init.log of netbios name
433 * Fix a DNLC panic when a UTF-16 string that cannot be
434 converted to a utf-8 string is provided as a filename.
436 * In the NSIS installer, delete old afslogon.dll/afscpcc.exe
437 from %windir%\system32 and fix the WinLogon\Notify\AfsLogon
438 location to refer directly to the new location in
439 %programfiles%\openafs\client\program
441 * Be more aggressive about responding to Netbios
442 NRC_BRIDGE/NRC_NOWILD errors. Rebind the Netbios name to
443 the LANA within the listener thread without performing a
446 * Use the WNet Enumeration of all connected drives as a filter
447 in the pioctl GetIoctlHandle operation. This should remove
448 the last situations in which there was a delay when opening
449 an Explorer Shell properties menu with the AFS Shell
452 * Fix the parsing of CellServDB in Windows to handle linked
453 cells as is done on Unix.
455 * Add support for linked cells to the cache manager,
456 netidmgr afs provider, and aklog. When two cells are
457 linked, a volume lookup that fails in one cell is looked
458 up in the other. This can be used for construction of
459 test cells that override specific volumes; cell renaming;
462 * Increase the length of cell and realm names that can
463 be entered in the NIM AFS provider.
465 * Actively prevent a 32-bit afsd_service.exe from
466 executing in the WOW64 environment.
468 * Protect against invalid IP address values in CellServDB
473 * Fix an Rx panic caused by an attempt to double free
474 an rx_packet when an error occurs while writing the
477 * Fix an exception cause by a race condition when
478 processing multiple file change notifications registered
479 by the Explorer Shell or other applications.
481 * Modify the Rx debug interface to use var_args and
482 permit Rx debugging to be used on 64-bit Windows
483 which previously had it disabled.
487 * Automated lock hierarchy enforcement added to the
488 osi base lock package. This permits detection of lock
489 violations that might but often does not result in
490 deadlocks. Enabled by default in the checked builds.
491 May be activated in release builds via the registry
492 value "LockOrderValidation" in the TransarcAFSDaemon
495 When a violation is detection an assertion is thrown
496 generating a dump file and a message indicating which
497 locks were obtained out of order is logged to the
498 Windows Application Event Log.
500 * Attempt to ensure that a dump file will be generated
501 whenever the service panics.
503 * Removed unused lock cm_bufGetMutex
505 * Many out of order lock acquisitions were detected
506 using the lock order verifier and were corrected.
508 * A race condition within the function path cm_GetSCache() ->
509 cm_GetNewSCache() -> cm_RecycleSCache() permitted a
510 cm_scache_t object to be issued simultaneously to two
511 threads. This would eventually result in a panic due
512 to the resulting undercount.
514 * Fix interpretation of the empty string as the ioctl path
515 to mean the current directory. "fs lsm", "symlink list",
516 etc. now return a "not a ..." error instead of "does not
519 * Add SMB_STRF_SRCNULTERM flag to smb_ParseStringXXX functions.
520 to indicate that the input string is a nul terminated string.
521 Assign ti when input strings are nul terminated instead of
522 length counted. (from 1.5.50)
524 * The CIFS protocol specification for handling NT_TRANSACT_CREATE
525 does not match the observed behavor. The 'nameLength' is specified
526 in bytes, not in characters. Fix the implementation to match
527 the observed behavior. (1.5.50)
529 * The cm_HaveAccessRights() attempt at deadlock avoidance by
530 calling lock_TryRead() on the parent directory cm_scache_t rw-lock
531 does not avoid the deadlock. Avoid the deadlock by enforcing the
532 lock order of the lowest vnode first. Then remove the infinite
533 loop avoidance in cm_SyncOp() that was returning an unwarranted
536 * File Server lock synchronization was not propertly enforced in
537 smb_ReceiveNTCreateX() and smb_ReceiveNTTranCreate().
539 * Another change to the short name generator. Plus and equal are
540 not valid characters in a short name according to the
541 RtlIsNameLegalDOS8Dot3() in kernel mode. They have been replaced
542 with left and right brace in order to minimize the impact.
544 * Fix a race condition that can result in a heap overwrite error if
545 the number of known file servers or vldb servers increases during
546 the preparation phase of a cm_CheckServers() operation.
548 * If the SMB Authentication would fail, a LSA allocated memory
549 buffer would be leaked.
551 * Rewrite the cm_cell_t allocation code to permit cell vldb server
552 lookups to fail without leaking cm_cell_t objects.
554 * Add "fs setcrypt auth" to permit the use of rxkad authentication
555 only mode. Not documented.
557 * Initialize Rx mutex and lock objects once and only once as part of
558 the DLL load process.
560 * When reporting the Volume Name "AFS" do not include the terminating
561 NUL in the reported length.
563 * Refresh Volume Location information upon receipt of a VNOVOL error
564 from the file server and avoid a race condition that prevented the
565 updated volume location information from being destroyed immediately
566 after it was acquired.
568 * In response to VNOVOL and VMOVED, remove the volume id from the server's
569 volume list so that the volume's up and down status is not affected
570 when the machine is shutdown.
572 * Many fixes to the Rx stack related to packet allocation and deallocation;
573 Rx debugging; Rx packet tracking across queues; and Rx packet loss caused
574 by overwriting the rx_call currentPacket pointer when it was already
575 referring to a packet.
577 * Rx Jumbograms are now off by default. To re-enable, add "RxJumbo" = "1"
578 in the TransarcAFSDeamon/Parameters key.
580 * "fs flush", "fs flushvolume", and "fs flushall" now destroy locally
581 constructed directory B+ trees for flushed directory objects.
583 * When cached pages for directory objects have been locally modified, do
584 not permit the locally modified pages to be mixed with pages subsequently
585 obtained from the file server. Doing so can result in directory
588 * If Ubix errors are received in response to VL_xxx operations record
589 the error on the server instance and permit the operation request to
590 be retried in case another server is able to answer the query. We have
591 seen at least one instance of a cell in which a VL server returned
592 UNOQUORUM (possible database truncation) when the other servers were
593 responding successfully.
596 * Optimize the assignment of write errors to dirty
597 buffers by applying the first error to all buffers
598 belonging to the same object instead of trying each
601 * Add resource version info to xstat_fs_test and xstat_cm_test
603 * Fix a delete vs delete[] usage problem in the WiX custom
606 * When creating a mount point, validate the volume name to
609 * When processing a mount point, pay attention to VL_BADNAME
612 * Remove a race condition from cm_volume_t recycling that could
613 have resulted in two threads making use of the same object at
616 * Fix a reference count leak on smb_vc_t objects introduced in
617 1.5.50. Permit the objects to be freed once again.
619 * Fix the output of "fs memdump" for smb_vc_t objects that was
622 * Add routines for debugging smb refcount errors.
624 * When shutting down the cache manager, do not permit AFS
625 callback requests to be processed. Doing so can result in
626 invalid memory accesses.
628 * Do not recycle a deleted cm_scache_t object unless the
629 refCount is 0. Doing so can result in a deadlock and
630 invalid memory access.
632 * Minimize the number of RPCs sent to the file server for
633 directory bulk status requests. Only request status for
634 objects that we are actually reporting on and only if we
635 do not already have status for the object and then make
636 sure that we return as much as we can at once via the SMB
637 interface but not so much that it exceeds the RDR timeout
639 * Enable bos restricted operations
641 * Fix the creation of submounts used by the afscreds and afs_config
644 * Avoid repeated retries when attempting to obtain access rights.
645 If the access rights were obtained successfully but do not
646 satisfy the request, fail the request.
648 * Fix error code reported when attempting delete a file on a
649 readonly volume or one that is marked with the readonly DOS
652 * Fix a short name truncation bug.
654 * Fix a heap corruption error when reading the CellServDB file
657 * Fix directory listings when the directory has "lookup" permission
658 only and the directory has never been previously accessed.
660 * Fix several tracking issues for cm_scache_t rw locks.
662 * Ensure that all buffers used to store cell names are the
665 * Restore Windows 2000 support (broken in 1.5.50).
666 LOCALE_INVARIANT was introduced in XP.
668 * Add RxUdpBufSize registry value. Default is 256K.
670 * Avoid multiple deadlocks between cm_freelanceLock and cm_scacheLock
672 * Do not include trailing NULs when returning file names in
673 a directory search response.
675 * Prevent RX from crashing the afsd_service by attempting
676 to allocate to many buffers at once when memory is fragmented.
678 * Avoid a deadlock when freeing cm_server_t objects
680 * Optimize search cell file. Do not search to end of file
681 if the cell has already been found.
683 * Remove explorer shell extension delay on local disks.
685 * Automatically pre-allocate 64 Rx packets per thread
687 * Add smb lock requests and stat cache lock allocations to
688 the output from fs memdump.
691 * Fix Large File Support (broken in 1.5.50)
693 * Fix "fs setacl" (broken in 1.5.50)
695 * Fix pioctl path evaluation of drive letter
696 mapped paths. (broken in 1.5.50)
698 * Fix installation of VC8 Runtime library in the EXE
699 Retail installer. (broken in 1.5.50)
701 * Avoid a potential deadlock during Volume Location
702 updates if all the file servers are unreachable at
703 the time the VL update succeeds.
705 * Ensure that rx connection object is references
708 * Fix propagation of over quota errors during cm_FSync()
711 Since 1.5.39 [1.5.50 released 16 July 2008]
712 * There were no releases between 1.5.39 and 1.5.50.
714 * FIRST RELEASE WITH UNICODE SUPPORT
716 * Pioctl interfaces to the cache manager refactored
717 to provide layering between the SMB specific code
718 and the general purpose ioctl operation.
720 * Garbage collect dead SMB virtual circuits as soon
721 as they are no longer being referenced. This avoids
722 problems with outstanding locks not being dropped
723 when the virtual circuit becomes invalid.
725 * Remove the IBM Administration Reference documentation
726 and replace it with the OpenAFS Command Reference
729 * Avoid calling rx_SetDeadTime and rx_SetHardDeadTime
730 functions each time a connection is about to be used.
731 Do not hold a lock on the rx connection object while
732 it is being selected. This avoids a race between
733 threads attempting to set the timeout values and
734 removes a bottleneck that was hampering performance.
736 * Ensure that the smb directory attribute is set
737 for all directory objects.
739 * Add vs2008 support to the NSIS installer scripts
741 * Replace the VC Runtime EXE installer with the MSI
742 installer in the NSIS installer scripts
744 * Properly delete the VC7.1 C Runtime libraries.
746 Since 1.5.36 [1.5.39 released 23 June 2008]
747 * There were no 1.5.37 or 1.5.38 releases for Windows
749 * Use Visual Studio version of vsprintf() instead of
750 home grown version that resulted in linking conflicts
752 * Prevent buserver, ptserver and fileserver from crashing
753 due to non-portable strftime() format strings.
755 * Prevent server crashes caused by rx_InitMorePackets()
756 being called prior to allocation of all of the required
757 synchronization objects.
759 * Fixed a memory leak in the internal B+ tree directory.
760 For objects with non-8.3 names, the 'longname' memory
761 allocation would be leaked during object deletion or
764 * Fixed a memory leak in the background daemon cm_CheckServer()
765 routine. The memory allocated to store the server list
768 * Obtain a missing lock around a call to cm_RemoveSCacheFromHashTable().
770 * Correct an abstraction layer violation. cm_scache_t objects
771 should be marked deleted in cm_Unlink() and cm_RemoveDir() and not
772 in smb_CloseFID(). Cleanup of deleted cm_scache_t objects should be
773 performed in cm_ReleaseSCache() when the reference count hits zero.
775 * Prototype cm_AdjustScacheLRU() and re-implement it using osi_QAddH().
777 * Do not reference 'smb_logp' in cm_vnodeops.c. Use 'afsd_logp'
780 * Ensure that scache object 'nextp' pointers are set to NULL
781 when the objects are removed from the hash table linked list.
783 * If a dirty buffer cannot be written to the file server because
784 of an access denied error, mark the error in the cm_buf_t
785 structure so that it can be dealt with instead of retrying
789 * Update CellServDB file to GCO Public 23 Apr 2008
791 * Fix a cm_buf_t reference count leak when attempts to write
792 dirty buffers to the file server from within cm_IncrSyncer()
795 * Prevent udebug from crashing.
797 * Activate RX Idle Timeouts. If the file server is busy for more
798 then 30 seconds, attempt to failover to another server without
799 marking the busy server down.
801 * Another VNOVNODE issue fixed. When writing a dirty buffer
802 to the file server, if VNOVNODE is received, mark all buffers
803 as invalid without further attempts to contact the file server.
805 * Improved performance on high latency links.
808 * Fix a bug preventing the re-initialization of the Freelance
809 root.cell contents when a change is made.
811 * Properly return CM_ERROR_NOSUCHFILE when removing a symlink
812 or mountpoint that does not exist.
814 * aklog, afscreds, integrated logon, and the NIM AFS credential
815 provider will all attempt to first acquire an afs service ticket
816 using the service principal: afs/<cell>@<CLIENT_REALM>. This
817 permits the proper detection of the cell realm when multiple
818 realms are local realms for the cell.
820 * Fix fs commands that re-use file identifiers instead of forcing
821 a path evaluation each time. This was broken in 1.5.33 when
822 hashes were added to the cm_fid_t structure.
824 * Fix symlink and mountpoint removal and then recreation when
825 applied to the fake Freelance root.afs volume.
827 * Fix SMB change notifications for the Freelance root.afs volume
828 when creating or removing symlinks and mount points.
830 * Force DNS AFSDB cell name lookups to lower case if the DNS
831 response is not all lower case.
833 * 32-bit installers are now built with Visual Studio 2005
835 * All vos.exe commands now support the "-noresolve" option which
836 forces the display of IP addresses instead of DNS hostnames.
838 * No longer hold the cm_volume_t mutex across VL_xxx RPCs. This
839 permits operations that do not require up to date volume
840 location data to continue while waiting for the RPC to complete.
841 Multiple threads waiting for a volume location update on the
842 same volume set no longer trigger multiple calls VL_xxx RPCs.
844 * Volume server references could be marked offline and never
845 reset to the not_busy state.
847 * Further optimizations to cm_Analyze() retry logic for
848 CM_ERROR_OFFLINE and CM_ERROR_ALLBUSY error handling.
851 * Optimize the DNLC by applying Interlocked incrementing for
852 statistics gathering and enabling greater use of read
853 locks instead of write locks
855 * Further optimize osi_Log operations when logging is disabled
856 by removing the need for the function call overhead.
858 * Add a new lock operation, Convert Read to Write, and apply
859 it throughout the cache manager. This function provides a
860 fast transition from a read lock to a write lock when the
861 caller is the only reader.
863 * Fix a timeout problem when opening files or request locks
864 when the file in question is already write-locked by another
867 * Fix a deadlock that can occur if "fs flushall" is executed
868 while applications are writing to AFS.
870 * The cache manager would re-read data from the file server
871 after file truncation when obtaining buffers to write to the
872 same file until the first StoreData RPC completes.
874 * Fixes error handling in cm_NTCheckDelete()
876 * Replaces cm_scache_t mutex with a read-write lock permitting
877 additional parallel access.
879 * Prevent an cm_scache_t reference undercount when evaluating
880 symlinks containing @sys when none of the active sysnames
881 match any of the available targets.
883 * Prevent the leak of a cm_scache_t rwlock if associated symlink
886 * Set RxMaxMTU registry value to 0 which is the equivalent of
887 it not being set at all.
889 * Reduce contention between BkgDaemon threads
891 * Prevent CM_SCACHEFLAG_ASYNCSTORE from being reset on a write
894 * Convert reference counts on cm_volume_t, cm_conn_t, cm_server_t
895 objects to use the faster Interlocked operations which avoid
896 the need to hold write locks.
898 * Permit rxkad errors other than RXKADEXPIRED to be treated as a
899 non-fatal error that forces a retry against another service
902 * Fix a crash in smb_WriteData that would occur if an application
903 opened the file as read-only and then sent a write request anyway.
906 * Respond to VL_NOENT errors by removing cm_volume_t object and
907 setting it to be the first object to be recycled. This is not
908 only a more efficient use of resources but it also prevents
909 repeated attempts to query the VLDB server for the volume from
910 the CheckOfflineVolumes routine.
912 * Change the defaults for RxEnablePeerStats and RxEnableProcessStats
915 * A first cut at a cache manager statistics monitor that can be used
916 to determine the necessary cache parameters to support the
917 application working set.
919 Off by default, the performance package can be activated by setting
920 "daemonPerformanceTuningInterval" in the service Parameters key.
921 As with the other daemon interval values the unit is in seconds.
923 At service start and each succeeding interval the cache manager
924 will write statistics to %TEMP%\afsd_performance.txt showing the
925 relative usage of cm_scache_t, cm_volume_t and cm_buf_t objects.
926 The FID statistics keep track of all FIDs seen by the cache manager
927 during the service session whether or not they are backed by any
928 live object in the cache.
930 These statistics are not stored in the cache file.
932 * Fix windows event logging broken in 1.5.33.
934 * Eliminate multiple calls to syscfg_GetIFInfo() which is used to
935 populate the IP address list for WhoAreYou/TellMeAboutYourself RPCs
936 as well as when computing the server preference list.
938 * When merging status after a StoreData operation prevent data
939 buffers from being discarded that we would prefer to keep.
940 (Bug introduced in 1.5.33.)
942 * Modify installers the set RxMaxMTU to 0.
944 * In the AFS Configuration Panel on the Advanced->Miscellaneous
945 dialog, the Background Daemons and Service Threads fields were
948 * Increase default number of Background Daemons to 4.
950 * Remove cm_volume_t reference from the cm_scache_t object. Permit
951 cm_volume_t objects to be recycled.
953 * Windows Error Reporting has shown a number of previously unseen
954 stack corruption errors in the Explorer Shell extension. The
955 minidump appeared to indicate the problem was in ParseAcl().
956 Reviewing the function it appears that numerous things could go
957 wrong if invalid input was provided. Added error checking and
958 data validation. Hopefully this will address the problem.
960 * Windows Error Reporting has shown crashes in aklog as a result
961 of krb5_cc_get_default() failures not being handled. They are
965 * The Rx library used a 32-bit type for sockets which was
966 truncating the socket value on 64-bit Windows. This
967 was not an issue in general but became an issue when
968 Rx "hot thread" support was enabled.
970 * Enable "Rx hot thread" support which ensures that while
971 a received message is being processed that a new thread
972 is allocated to receive the next message to arrive.
974 * Increased parallel access to smb_rctLock protected
975 resources. Reduce the number of times that the
976 lock is dropped and re-obtained.
978 * Increased parallel access to cm_scache_t and cm_buf_t
979 resources by replacing the previous global write lock
980 protected reference counting mechanism with one built
981 upon the use of InterlockedIncrement and InterlockedDecrement.
983 * Implement a hash for cm_fid_t to reduce comparison costs
984 and inline the cm_FidCmp() function.
986 * Redefine the BUF_HASH and BUF_FILEHASH in terms of the cm_fid_t
987 hash which has a better distribution
989 * Modify cm_ConsiderPrefetch to evaluate the amount of data in
990 the most recent read request instead of the next chunkSize.
991 cm_chunkSize can be dozens or hundreds of buffers. As a result
992 too much time is spent performing the evaluation.
994 * Fix the usage of cm_scache_t bufCreateLock. The purpose of this
995 lock is to prevent the creation of new buffers while a truncation
996 is being performed. All references to bufCreateLock have been
997 removed except in two places:
998 i. a write-lock surrounding the function that calls buf_Truncate()
999 ii. a read-lock within buf_GetNewLocked() that actually allocates
1002 * Modify the CM_CONFIG_DATA_MAGIC value to include a
1003 CM_CONFIG_DATA_VERSION number which value be used to force the
1004 replacement of the cache file contents when incompatible changes
1005 are made between releases.
1007 * CM_SCACHESYNC_ASYNCSTORE should not be ordered by
1008 cm_SyncOpCheckContinue
1010 * Avoid calls to multi_Rx if nconns == 0
1012 * Modify smb_WriteData to perform background writes based upon the
1013 crossing of 'smb_AsyncStoreSize' boundaries instead of cm_chunkSize
1014 boundaries. This will slow down writes from the SMB interface but
1015 will avoid the risk of the CIFS client disconnecting from the AFS
1016 client SMB server. The default value for 'smb_AsyncStoreSize'
1019 * SMB AsyncStore functionality is on by default but can be
1020 disabled using the registry value:
1022 HKLM\SOFTWARE\OpenAFS\Client DWORD "EnableSMBAsyncStore"
1024 smb_AsyncStoreSize can be adjusted with
1026 HKLM\SOFTWARE\OpenAFS\Client DWORD "SMBAsyncStoreSize"
1028 The value must be a multiple of the buffer block size
1029 and cannot be larger than the chunk size.
1031 * Do not leak a read lock on the bufCreateLock
1033 * Do not forget to call cm_SyncOpDone if cm_SyncOp succeeded
1036 * In MergeStatus, only remove clean buffers from the hash table.
1038 * Avoid the need to update the data version number on each buffer
1039 associated with a scache when MergeStatus is called after a
1040 StoreData by maintaining a range of valid data versions as part
1041 of the cm_scache_t object.
1043 * Profiling shows large numbers of blocked calls in cm_HoldSCache
1044 from cm_HaveAccess... We can safely avoid the call so do so.
1046 * Add "ForceAnonVLDB" service parameters registry value. Default
1047 is 0. Set to 1 to force the use of anonymous connections to
1048 the VLDB servers. This is useful when using krb5 tokens and the
1049 VLDB servers have not been updated.
1052 * Do not attempt to search for a cell name that matches one
1053 of the special share names "svrsvc", "wkssvc", "ipc$"
1055 * Issue permission denied warnings from afscreds if attempts
1056 to modify the startup shortcut or registry values fail.
1057 This happens under Vista UAC.
1059 * Do not return BAD_NETWORK_ERROR in response to attempts
1060 to create directories or files in the Freelance root.afs
1061 volume. Instead return ACCESS_DENIED.
1063 BAD_NETWORK_ERROR was being sent because the Freelance cell
1064 has no vldb servers. Short circuit the error by testing for
1065 freelance in CreateFile and MakeDir
1067 * Treat srvsvc, wkssvc and ipc$ as pseudo files. Permit them
1068 to be opened and closed without actually existing. For now
1069 we treat them as equivalent to ioctls but this should
1072 * Fix RAP processing to not reject names that exist in the
1073 root.afs volume for GetInfo queries.
1075 * In afscreds, use GetUserNameEx to obtain the username instead
1076 of searching the registry. Try UserPrincipalName first and
1077 fallback to SamCompatible.
1079 * Stop linking to rpcns4.dll which is no longer supported on
1080 Vista and 2008. TaAfsAdmSvr is going to have to be
1083 * When processing dfs referral requests, if the sharename is
1084 a partial match and therefore will not be added automatically
1085 to the share list, make sure that we return an error.
1087 * Avoid a crash when logging VNOVOL errors returned from
1088 cm_CheckOfflineVolume.
1090 * When trace logging is disabled, do not permit osi_LogSaveString
1091 to perform the string copying. By disabling the string
1092 copies, throughput of the service more than doubles.
1094 * when checking offline volumes, update the cell vlserver info
1095 if required and don't perform a check if there are no servers
1098 * Add volumes in the vl_unknown state to the list of objects to be
1099 checked as offline or retried when an allbusy error is received
1102 * Add support for configuring rx_ExtraPackets
1104 * All calls to buf_Get() must be protected by a read lock on the
1105 cm_scache_t bufCreateLock
1107 * When a MergeStatus results in the invalidation of buffers, do
1108 not remove buffers from the hash tables that have active references.
1110 * Upon receiving a VNOVNODE error from a file server, be sure to
1111 invalidate any file locks. This prevents on-going attempts to
1112 extend the locks and will cause subsequent attempts to access
1113 the file to receive a invalid handle error.
1115 * Prevent smb_ReceiveTran2SearchDir from adding terminating NULs
1116 to directory entry names.
1118 * No longer discard tokens when RXKADBADTICKET error is received.
1119 Instead return STATUS_NO_KERB_KEY error.
1122 * Add "cmdebug <host> -cellservdb" option and client service
1123 implementation permits querying the vldb server for each
1124 cell known to the cache manager. The output is in a form
1125 that permits it to be used as a CellServDB file.
1127 * Add Symlinks->Show option to the AFS Explorer Shell Context
1130 * Prevent buffer overrun in the cmd parser used by command
1131 line tools such as fs, vos, pts, etc.
1133 * The path ioctl operations have several issues:
1135 (1) the specified path for "fs examine, whereis, whichcell,
1136 flush" and so always has the follow symlinks and mount
1137 points semantics. This makes it impossible to determine
1138 what the FID of a symlink or mount point is.
1140 (2) "fs examine" out is not the result of a single pioctl
1141 operation but is actually the combined output of half
1142 a dozen operations. Path evaluation is an expensive
1143 operation. It would be faster if the caller could
1144 evaluate the FID first and then perform all of the
1145 rest of the operations by specifying the FID instead
1148 (3) fs output reports all objects as files. By adding a
1149 GetFileType pioctl more informative output can be provided
1150 that indicates what type of object the path evaluates to.
1152 (4) the Windows fs command includes a number of commands that
1153 do nothing but exist only because the Unix cache manager
1156 A new extendible data structure cm_ioctl_query_opts_t has been
1157 added which can be optionally specified with pioctls that do not
1158 already require input data. The first two fields of this structure
1159 are 'literal' and 'fid'. The literal field is used to indicate
1160 whether the last component of the path should be evaluated following
1161 symlinks and mount points.
1163 The fid field permits a fid to be specified.
1165 * A new GetFileType pioctl has been added. The type of objects are
1168 * A new "-literal" option is available for "fs examine, flush, whereis,
1171 * Unimplemented fs commands have been removed.
1173 * There are circumstances where a volume object is being accessed and
1174 the volume is marked indicating that the volume location information
1175 is out of date but where it is also pointless and perhaps dangerous
1176 to block waiting for the rpc to complete. One example is when
1177 processing the cmdebug requests. If we know that we are not going
1178 to use the volume object to contact a server, then we can now set
1179 the CM_GETVOL_FLAG_NO_RESET flag.
1181 * when moving up in the directory tree we search the recorded fid list
1182 to find a matching fid that we have already crossed. we must also
1183 reset the fid count based upon what we discover.
1185 * Cell names are published as share names. As a result they are searched
1186 for as part of the Dfs Referral evaluation. If share "foo" can not be
1187 reached, the CIFS client will fallback to searching for "fo" as well.
1189 Since the freelance client automatically adds a symlink for prefixes
1190 of the cell name, this results in both "foo" and "fo" being added to
1191 the freelance root.afs volume. This patch prevents that negative side
1194 * when tracking Kerberos credential cache names be sure to include
1195 the cache type prefix.
1197 * Microsoft has assigned OpenAFS a network type value
1199 * The algorithms that were used to produce a Netbios name were broken.
1200 The name that was produced when the hostname was too long was in fact
1201 longer than the maximum permitted netbios name.
1203 Also, the "NetbiosName" value was not used as a suffix when the loopback
1204 adapter was not installed. The hardcoded string "AFS" was used instead.
1206 * avoid another deadlock during server probes initiated by the ipaddr
1207 change daemon thread.
1209 * prevent the afs client service from crashing when fs newcell is
1212 * In the NSIS installer, move afslogon.dll, afscpcc.exe and afs_cpa.cpl
1213 to \Program Files\OpenAFS\Client\Program in order to get them out
1214 of the %windir% tree
1216 * EDQUOT == WSAEDQUOT. Define it in the right places so that in the
1217 end STATUS_OUT_OF_QUOTA can be returned to the application.
1219 * Modify the search order for determining the location of CellServDB
1220 and other client configuration files.
1222 1. AFSCONF environment variable
1224 3. NEW - use All Users\AppData\OpenAFS\Client only if CellServDB exists
1225 4. use Program Files\OpenAFS\Client
1227 * More Vista Power Management tweaks
1229 * Close a socket handle leak in the kauth routines
1231 * Don't hold a cm_volume_t mutex across RPCs
1233 * Convert downTimes to use clock time instead of relative times.
1234 This permits correct comparisons with expiration times.
1236 * Update the Wix installer to properly install the prerequisites
1237 for translate_et.exe
1239 * Do not translate pioctl paths as they are not converted by the
1242 * Implement multi_rx probes for checking whether servers are up
1243 or down. Turns probes into a constant time activity instead
1244 of one dependent upon the number of known servers.
1246 * Treat VNOVOL errors the same as VOFFLINE. Do not force an
1247 update of the volume location data and retry. VNOVOL means
1248 that the volume could not be attached. It doesn't mean that
1249 the volume is not on this server.
1251 * Modify the Explorer Shell AFS Context menu Symlink->Add to
1252 not attempt to validate the target path. There is no reliable
1255 * Improve performance of \\afs\<cell><mount-type><volume>\
1256 access by eliminating extraneous attempts to search the
1257 CellServDB file and DNS AFSDB records. Share names
1258 containing '%' or '#' are not valid cell names.
1260 * Return STATUS_RANGE_NOT_LOCKED if the application attempts
1261 to unlock a range that is not currently locked.
1263 * Enable the 32-bit tools to read the 64-bit registry keys
1264 when executing in the WOW64 environment.
1268 * Fix buffer refcount leak introduced in 1.5.29
1270 * Prevent the NIM AFS Provider from crashing when configured to use
1271 Kerberos v4 for token acquisition in situations where no Kerberos v4
1272 support is available. For example, 64-bit KFW or sites that delete
1273 the Kerberos v4 DLL.
1275 * In afsd_init.log, support logging cache sizes greater than 2GB.
1277 * The volume status output of "fs examine" was being determined based
1278 upon the wrong error value. Instead of using the pioctl() return
1279 code, the errno value must be used.
1281 * Prevent invalid pts auto-registration attempts by aklog and the
1282 NIM afs provider when Kerberos referrals are in use. (krb5 1.6+)
1284 * Add synchronization protection to all of the SMB Listener State
1285 variables. This prevents race conditions when the addition or
1286 loss of a network adapter takes place.
1288 * On Vista, there is a race condition between the restart of the
1289 afs client service and the network adapters. Ensure that the
1290 afs client service can handle switching between loopback and
1291 non-loopback modes of operation.
1293 * Fix FollowBackupPath to work with mount points that are not
1294 explicitly read/write.
1296 * Fix a deadlock that has been observed on Vista while resuming
1297 after at least four hours of sleep.
1299 * Add new HKLM registry configuration options that can be used
1300 to provide configuration hints to the NIM AFS Provider when
1301 creating new identities. See release notes for details on
1302 HKLM\SOFTWARE\OpenAFS\Client\Realms\.
1305 * Speed up write buffer operations by holding a global read-lock
1306 instead of a write-lock when updating the buffer data version
1307 number. The data version is protect by the associated stat
1310 * Speed up read buffer operations by removing out of date buffers
1311 from the buffer hash tables.
1313 * Protect against a null pointer dereference in afslogon.dll
1314 if QueryAdHomePathFromSid() fails.
1316 * Remove a race condition while accessing private freelance data
1319 * Update the buffer data versions when Freelance directory data
1322 * Complete the implementation of smb_ReceiveTran2GetDFSReferral().
1325 * Return EAS_NOT_SUPPORTED in response to EA read/store/query
1326 instead of returning zero length.
1328 * Return ERROR_RETRY when a server sends VBUSY.
1330 * Return PATH_NOT_COVERED instead of BADSHARENAME for DfsLinks when
1331 a volstat plug-in is in use.
1333 * Enhance the volstat interface to support Dfs path mappings.
1335 * Fix "symlink list" to do the right thing for Dfs links
1337 * Fix callback expiration that was broken in 1.5.27.
1339 * Fix a crash at shutdown caused by the lack of synchronization
1340 within the daemon threads (cm_Daemon, cm_BkgDaemon, cm_IPAddrDaemon)
1342 * Hold the cm_server_t mutex while walking or modifying the
1343 cm_server_vols_t list in order to avoid a race condition.
1345 * Ensure that the cm_server_t refCount is safely modified.
1347 * The pthread library implementations of pthread_cond_wait
1348 and pthread_cond_timedwait were failing to return with
1349 the mutex held when error conditions occurred. This could
1350 result in race conditions wherever they are used. One case
1351 in which problems were seen was the Rx event_handler().
1353 * Prevent cm_GetBuffer() from locking all buffers associated with
1354 a stat cache object before deciding if the current buffer is up
1357 * Fix cm_BkgPrefetch to actually ensure that there is a cm_buf_t
1358 object for each offset that we will attempt to reserve.
1360 * Reduce the default number of background daemon threads used to
1361 prefetch or store to one in order to prevent two or more threads
1362 from stepping on each other.
1364 * Reduce the default chunk size to 256KB in order to prevent read
1365 timeouts on low bandwidth connections.
1367 * Fix a race condition that could result in threads becoming stranded
1368 when synchronizing multiple requests on the same stat cache object.
1371 * The GiveUpAllCallBacks functionality that was added in 1.5.21
1372 has been disabled. It has been disabled because it turns out
1373 that the RPC is improperly implemented in all file servers
1374 previous to 1.4.6. The RPC is executed without the proper
1375 locks being held resulting in data corruption and eventual file
1376 server failure. The functionality will be re-enabled in a
1379 * a couple of .readonly Volume Callback optimizations were
1381 - Apply most recent volume callback to all stat cache objects
1382 in the volume. This avoids premature callback expirations
1383 and unnecessary FetchStatus calls.
1384 - Add option to permit attempts to renew .readonly Volume
1385 Callbacks on a periodic basis to avoid invalidation of stat
1386 cache data. [Requires daemonCheckVolCBInterval be set to
1387 non-zero number of seconds. 1800 (30 minutes) is suggested.]
1389 * Adds support for 64-bit data version values. Previous releases
1390 only supported 32-bit values. Larger values sent by the file server
1391 would result in rollover.
1393 * aklog and asetkey no longer generate an exception if Kerberos for
1394 Windows is not installed.
1396 * If the LSA authentication functions report a temporary out of memory
1397 condition, fail the current authentication and do not cause the AFS
1398 client service to terminate unexpectedly.
1400 * When the AFS client service is shutting down, prevent the daemon check
1401 thread from executing additional checks which might attempt to access
1402 discarded locks or unmapped memory.
1404 * Correct a double free condition in the NIM AFS credentials provider.
1406 * Add "FollowBackupPath" service parameter registry value.
1407 This provides equivalent functionality to the UNIX afsd -backuptree
1408 option. When set to a non-zero value, normal mount points originating
1409 in a .backup volume will prefer a .backup volume.
1412 * Fix NSIS uninstall from the .MSI installer.
1414 * Add Stack Frame logging for AMD64
1416 * Add support for EWOULDBLOCK to cm_Analyze. If the file
1417 server returns EWOULDBLOCK, retry the request every two
1418 seconds for up to the RDRtimeout.
1420 * Reorganize the order of the includes to ensure that EWOULDBLOCK
1421 is not assigned the same value as EIO.
1423 * Update CellServDB to GCO Public 25 Oct 2007
1425 * Add additional validation and error handling code after
1426 each call to getSlot(). If an invalid slot is returned,
1427 return NONODE. If the invalid slot is returned when
1428 extracting a data node, invalidate the tree.
1430 * Modify compareKeys() to always perform a case-insensitive
1431 comparison and only perform a case sensistive comparison
1432 if the case-insensitive one matches. This ensures the
1433 ordering is consistently reported.
1435 * Add lock assertions to ensure that all calls are being
1436 performed with the correct locks being held. There have been
1437 some crash reports that provide stack data that does not appear
1438 to be possible unless there is a race. However, there are no
1439 obvious locations where the race is taking place and the test
1440 suite indicates that all of the correct locks are being held.
1441 We shall see what happens in the field.
1443 * For consistency replace all calls to findKey in which the range i
1444 s (1,numentries) with calls to getSlot().
1446 * Optimize the depth search loop by testing the slot value in the
1447 for statement instead of forcing the loop to be broken later.
1449 * Reorganize the locking for cm_BeginDirOp and cm_EndDirOp.
1450 There are a number of locations where locks are obtained, dropped,
1451 and reobtained. This reorganization attempts to accomplish several
1454 (1) be optimistic for the most common case so it will be fast
1456 (2) add consistency checks after each location where locks are
1457 dropped and re-obtained. If we lose a race in cm_BeginDirOp
1458 and the bplus tree is out of date, retry until we get to a
1459 consistent state that we can use.
1461 (3) Ensure that all operations take place with the correct locks.
1463 * One of the issues that has become a serious problem since the
1464 addition of the local directory updates is that although cm_SyncOp
1465 synchronizes operations, it does not preserve the order of requests.
1466 This has always been a problem in that it has been possible for a
1467 request to fail to complete due to its worker thread's bad luck.
1468 When a request takes longer than the Windows SMB Redirector's timeout,
1469 the SMB Redirector tears down the SMB virtual circuit.
1471 When using the local directory updates it is really important that
1472 the directory update operations complete in the order that they were
1473 sent to the file server. If they don't, then the local directory state
1474 and the file server state will not match and the local directory state
1475 must be discarded which in turn forces a new read of the entire
1476 directory contents over the network.
1478 This patch adds a new cm_scache_waiter_t object that is used to store
1479 the current thread, buffer, and syncop flags within a waiters queue
1480 on each cm_scache_t object. If a thread is forced to sleep in
1481 cm_SyncOp, upon waking it will check to see if there are any other
1482 threads waiting that are attempting to perform a similar task ahead
1483 of it in the queue. If yes, the thread goes back to sleep. If not,
1484 it goes ahead and enters the cm_SyncOp conflict resolution block.
1486 This patch has the additional side effect of reducing the number of
1487 competing threads that must obtain the cm_scache_t mutex and process
1488 the cm_SyncOp conflict resolution block. As a result, the overall CPU
1489 utilization of the service and the clock time associated with processing
1490 requests will be reduced.
1492 * assert that the cm_scache_t mutex is held when calling
1493 buf_ForceDataVersion().
1495 * rename findNode to leafNode in bplus_Lookup
1497 * replace all OutputDebugString calls with osi_LogX calls
1499 * modify bestMatch to special case the return values for leaf nodes.
1500 If an entry is above or below the values available in the leaf node
1501 return BTLOWER or BTUPPER instead of BTERROR.
1503 * In insert and delete operations check for BTLOWER/BTUPPER and isleaf,
1504 if true convert to either slot 0 or Max and perform the insertion.
1505 This produces easier to read code when performing lookups.
1507 * modify lock_AssertXXX macros to call osi_assertx() and provide a
1508 descriptive message.
1510 * the difference between osi_assert() and osi_assertx() is whether or not
1511 a description is specified as part of the call. When no message is
1512 specified afsd_notifier() is called with a NULL msgp parameter. This
1513 results in a null pointer dereference during the EventLog() call.
1515 As a result, none of the information describing the assertion is written
1518 This commit sets a default message for use when no other message is
1519 provided and it replaces all calls to osi_assert with osi_assertx and
1522 * Migrate B+ tree search key into thread local storage
1525 * Fix an uninitialized variable in lana_IsLoopback() which
1526 could result in a random determination that a loopback
1527 adapter is not a loopback adapter. This would result in
1528 a computed Netbios service name of %MACHINE%-AFS instead
1529 of AFS. This is a serious problem if the value computed
1530 by the AFS Client Service and the client process pioctl
1531 calls produce different results as it will result in a
1532 communication failure.
1534 * Fix a crash in the B+ Tree code that results when a file
1535 rename only results in a change of case.
1537 * Fix the pioctl path parsing. The previous algorithm
1538 broke the current directory of the client into two parts,
1539 the TIDpath which represented the device root and the
1540 relative path. The problem is that this is not sufficient
1541 from proper processing of symlinks and mount points.
1543 What is required is that the last component of the path
1544 must not have the symlink or mount point be evaluated
1545 whereas symlink and mount point evaluation must be performed
1546 for the rest of the relative path. Therefore, the path is
1547 now split into three parts: TIDpath, intermediate path,
1550 "fs examine" now reports the data of the symlink or mount
1551 point object instead of the target and "symlink list" and
1552 "symlink remove" now works. The "Explorer Shell Extension"
1553 menu now is displayed for symlink objects.
1555 * More informative symlink.exe error messages.
1557 * When the cmd.exe "del" command is used to remove a symlink,
1558 not only was the link object removed, but the stat cache
1559 object for the target was marked deleted. This resulted
1560 in cache corruption which would require the destruction
1561 of the cache to correct.
1563 * Add undocumented registry value, "OfflineReadOnlyIsValid"
1564 which results in cached contents from read only volumes
1565 to be treated as valid when there is no callback registered
1566 and all servers are down or all instances are offline.
1568 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
1570 * Add documented registry value, "AcceptDottedPrincipalNames"
1571 which is used to permit the acquisition of AFS tokens
1572 using Kerberos v5 principal names which include a dot in
1573 the first component.
1575 {HKLM\HKCU}\Software\OpenAFS\Client
1577 * "fs uuid" does not require "AFS Client Administrator" group
1578 membership to display the current UUID value.
1581 * Update NSIS installer to version 2.30
1583 * Update Wix installer to version 2.0.5325
1585 * 64-bit KFW support:
1586 - Network Identity Manager plug-in
1588 - 64-bit krb5 enable afscreds.exe, afs server manager,
1591 * Enable B+ tree Directory Search by default.
1593 * NIM plug-in is now compatible with FILE: ccaches
1595 * Install instloop.exe into Client/Programs/ even though
1596 it isn't required. Permit it to be used by organizations
1597 that clone Windows systems.
1599 * On a suspend, only drop callbacks for servers that
1600 are online. If the server is offline, maintain the callback
1601 until the server is once again available.
1603 * Prevent against acquisition of a directory write lock when
1604 there are no updates to be performed. This was resulting
1605 in a write-lock leak which later on would lead to a deadlock.
1607 * Add configurable file pre-fetch mode. For files that match
1608 the specified file extensions, pre-fetch the entire file into
1611 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
1612 (MULTI_SZ) PrefetchExecutableExtensions
1614 * Fix the client_osi large integer support so that return values
1615 can be properly assigned to large integer variables.
1617 * Do not hold cm_scache_t mutex across cm_GetCell() calls
1618 since cm_GetCell() can block in a RPC.
1620 * Add short file name support to B+ tree searches.
1622 * Resolve a conflict between the desire to notify file servers
1623 that the client workstation is suspending and the desire to
1624 permit objects with valid callbacks when the client suspends
1625 to continue to use those objects from the cache if the client
1626 resumes without network connectivity.
1628 * Fix stack variables to ensure that all buffers storing paths
1629 can handle AFSPATHMAX (1024) characters. Paths of 256 characters
1630 would cause a stack overflow condition.
1632 * On Vista, force afs_config.exe to run with the highest privileges
1633 available to the end user. For a user in the Administrators
1634 group running under UAC, this will cause afs_config.exe to be
1635 be run with full Administrators privileges after the user
1636 approves the privilege escalation.
1638 * More VNOVNODE handling issues. If a dirty buffer cannot be
1639 written due to VNOVNODE, clear the buffer and ignore the error.
1640 If a cm_scache_t object has been marked deleted due to a VNOVNODE
1641 error, the next time the smb_fid_t object that references it is
1642 used return either NOSUCHFILE or NOSUCHPATH and close the
1643 smb_fid_t object. Subsequent attempts to use the smb_fid_t will
1644 result in an invalid handle error.
1646 * For Vista, set the RequestExecutionLevel of the NSIS installer
1649 * For the build environment, support the generation of manifests
1650 when using the CL1310 (VS .NET 2003) compiler/linker.
1652 * There was a bug in OutputDebugHexString() which could overflow
1653 a stack buffer resulting in unexpected process termination
1654 of afsd_service.exe if the input Security Token contained
1655 many bytes equivalent to the '%' char.
1657 * Add directory lookup statistics to "fs memdump" output.
1661 * Windows uses case-insensitive file name pattern matching
1662 but AFS is a case sensitive file system. The AFS3 directory
1663 format is block based, uses network byte order and
1664 includes a hash table for fast case sensitive lookups.
1665 This causes several problems for the Windows AFS client.
1666 (1) Traversing the directory blocks is cpu expensive
1667 (2) A hash table miss does not indicate that the desired
1668 entry does not exist.
1669 (3) Determining whether a non-ambiguous inexact match or
1670 the entry does not exist requires a linear traversal
1671 of the entire directory.
1672 These issues often result in 100% CPU utilization.
1674 These issues are addressed by building a modified B+ tree for
1675 each directory and then using the B+ tree for searches.
1677 This feature is disabled by default. To enable it set
1678 the "BPlusTrees" registry value to 0x01 in the AFS Client
1679 Service Parameters key.
1681 * Correct a null pointer dereference in the symlink recursion
1682 detection code. (Introduced in 1.5.22)
1684 * Add the Network Identity Manager AFS Provider to AMD64
1689 * A new registry value, "BlockSize", can now be used to adjust
1690 the size of the internal buffers used to store file data.
1691 The default is 4KB. When set, the values 1..1024 specify
1692 multiples of 4KB. Values > 1024 specify actual block sizes
1693 and must be multiples of 4KB. The actual block size in KB
1694 can not be larger than the chunk size which defaults to 1MB.
1696 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
1699 * pts.exe and ptserver.exe built with -DSUPERGROUPS
1701 * Service Manager shutdown hints have been added. The hint
1702 specifies that it may take the AFS client service up to two
1703 minutes to shutdown.
1705 * Fix the recursion detection feature introduced in 1.5.22.
1706 The previous implementation broke the evaluation of paths
1707 using ../path or ./path. It also broke the evaluation of
1708 symlinks to absolute paths. For example, /afs/cell/path.
1710 * Fix an uninitialized variable in cm_MakeDir that could result
1711 in crash if the requested directory could not be created.
1713 * Fix a the VolumeStatusNotification log messages to save
1714 the input string in the circular log buffer so garbarge
1715 won't be output to afsd.log when the log is dumped.
1718 * Increase the SMB Maximum Path length to 260 characters
1719 from 256 as that is what CIFS supports.
1721 * On 64-bit Windows, use a unique description string for
1722 the 32-bit AFS Shell Extension.
1726 * In the AFS Admin Library, replaced all of the
1727 Ubik_Call(XXX_RPC,...) calls with Ubik_XXX_RPC(...). This
1728 allows parameter checking and proper type conversions to be
1729 performed. This is crucial when passing parameters that are
1730 64-bit wide on systems in which 'long' is a 32-bit value.
1731 Otherwise, the parameter list on the stack is interpretted
1732 incorrectly and the function not only doesn't succeed but it
1733 might corrupt memory and if lucky crash the application.
1735 The only program that we care about that uses this library
1736 is the AFS Server Manager. Now it doesn't crash.
1738 * Fixed smb directory searches for non-wildcard names. The
1739 optimized function should not have set the bulk in progress
1740 flag. This would adversely affect later directory searches
1741 that required a bulk search.
1743 * Detect infinite recursions (up to 512 stat cache objects
1744 in a chain.) If an infinite recursion is detected, return
1745 an error to stop the CIFS client from querying us forever.
1747 * Executables executed out of AFS name space can have their
1748 pages swapped out by Windows. Those pages will be reloaded
1749 when they are required. However, if the cache manager no
1750 longer has a callback and all of the file servers with
1751 the volume data are down, then the application will crash
1752 due to the inability to read the required page.
1754 We now track the down time of each server. If the callback
1755 expires after the server that issued it went down and all
1756 servers are down, we delay the expiration to permit the data
1757 in the cache to be continue to be used by Windows or an
1760 * The Windows AFS client has always had very poor performance
1761 with regards to directory management. In the case of
1762 directory updates caused by create, rename,and delete operations,
1763 the AFS cache manager will now perform local directory updates
1764 when possible in order to avoid re-reading the directory from
1765 the file server. This will be done in whenever both of the
1766 following conditions are true:
1768 - the data version on the directory increased by one as a
1769 result of the operation
1771 - all of the directory buffers necessary to make the change
1772 are present in the cache
1774 * The error message in the AFS Control Panel describing the
1775 need to be a Windows Administrator has been replaced with
1776 a message describing the need to be in the AFS Admins Group.
1778 * If the vldb server list is empty when attempting to evaluate
1779 a cell mountpoint it was possible for a crash to occur. This
1780 can only happen if there is an existing mount point and the
1781 is a failure attempting to load the server list from the
1782 CellServDB file or via AFSDB queries.
1784 * dirty buffers only have their dirty bytes stored to the file
1785 server instead of the entire chunksize
1787 * the default chunksize has been increased from 128KB to 1MB
1792 * changed the enum values for cm_serverRef_t state info to use a
1793 private name space to avoid collisions (srv_)
1795 * added a srv_deleted state for cm_serverRef_t objects. This
1796 state is set when cm_FreeServerList() is called with the
1797 CM_FREESERVERLIST_DELETE flag set. cm_FreeServerList() may
1798 not always delete the cm_serverRef_t from the list if it is
1799 still in use by another thread. the srv_deleted state means
1800 the object's contents are no longer valid and it must be
1801 skipped. It will be deleted the next time the object is
1802 freed and the refcount hits zero.
1804 * the srv_deleted state is also used when a file server reports
1805 either VNOVOL or VMOVED instead of marking the cm_serverRef_t
1806 as offline. This is done to prevent additional usage of the
1807 stale vldb data while waiting for the update volume request
1810 * added a state field to the cm_volume_t object (enum volstate
1811 vl_ name space) that maintains the state of the volume based
1812 upon the states of all of the cm_serverRef_t and cm_server_t
1815 * modified cm_UpdateVolume() to set the state of the cm_volume_t
1816 RW, RO, and BK to either vl_alldown or vl_online. There can't
1817 be any other states because cm_UpdateVolume() destroys any
1818 previous knowledge we might have had regarding busy or offline
1821 * modified cm_UpdateVolume() to update the volume name in the
1822 cm_volume_t to the volume base name if the previous value was
1825 * modified cm_FollowMountPoint() to check to see if the volume
1826 name is a volume ID and if so call cm_GetVolumeByID instead
1827 of cm_GetVolumeByName. This ensures that volume IDs are always
1828 looked up as numeric values. There is no longer a need to
1829 maintain a separate cm_volume_t containing the string representation
1832 * Added a flags parameter to cm_GetVolumeByName() and cm_GetVolumeByID().
1833 The first flag is a "CREATE" flag which is set by all existing
1834 calls. The flag is not set by calls to cm_GetVolumeByID() from
1835 the server probe code when volume status is being updated. We
1836 do not want the server probe operation to result in additional
1837 turnover in the cached volume data. The second flag is NO_LRU_UPDATE
1838 which is set when the server probe code updates the volume status.
1839 This flag will be used to prevent the server probe operation from
1840 changing the order of the least recently used queue.
1842 * Modified cm_GetVolumeByName to ensure that only one cm_volume_t is
1843 allocated for a given set of normal, readonly, and backup volumes
1844 regardless of whether or not the volume is accessed via name or
1845 ID number. The cm_volume_t namep field is always the base name
1848 * Added a new volume state, vl_unknown. This state is used as
1849 the initial state for all cm_volume_t when the cache manager starts,
1850 for each cm_volume_t at creation, and for each cm_volume_t when
1851 recycling. The cache manager does not know the state of all
1852 volumes in the world, only those that are in the cache and for
1853 which it has queried the VLDB and hosting file servers.
1855 * modified cm_GetVolumeByName() to initialize the state of a
1856 volume to vl_unknown. The actual state will be set when a
1857 cm_VolumeUpdate() call completes successfully.
1859 * changed name of scache hash table variables to avoid ambiguity
1860 when adding hash tables for volumes
1862 * fix a buffer overrun in sys\pioctl_nt.c pioctl().
1864 * modified cm_UpdateVolume() to handle the case in which there is
1865 no RW volume but there is are RO volumes for a given base name.
1866 This is done by querying for the ".readonly" volume name if the
1867 base name does not exist in the VLDB. We never query for the
1868 .backup name because under the current usage model a .backup
1869 volume may only exist on the server that the read-write volume
1870 is located. If there is no RW volume, there can be no .backup.
1872 * Added four hash tables for cm_volume_t objects to improve the
1873 search time of cm_GetVolumeByID(), cm_GetVolumeByName() and
1874 cm_ForceUpdateVolume(). One each for Name, RWID, ROID, and
1875 BKID. Three ID hash tables are necessary as long as it is
1876 desireable to maintain a single cm_volume_t containing all
1877 of the related RW, RO, and BK volume data. Having the RW and
1878 RO volume data in the same object is necessary for the
1879 implementation of cm_GetROVolumeID() which returns either the
1880 RO or RW ID depending upon the existence of RO volume instances.
1882 * Added a volume LRU queue so that volume reuse becomes fairer.
1883 This does not replace the all Volumes list which is used when
1884 it is desireable to walk a list of all the volumes whose order
1885 is not going to change out from underneath you which makes it
1886 safe to drop the cm_volumeLock.
1888 * handles volume hash table updates where volume name to
1889 volume ID number changes. The volume name remains
1890 constant in the cm_volume_t. if a vos rename is performed,
1891 the name of the volume will change and the volume IDs will be
1892 updated. Subsequent access to the old volume ID will create a
1893 new cm_volume_t with the new name.
1895 * Added a daemon thread operation to query the state of volumes
1896 listed as busy or offline. cm_CheckBusyVolumes() calls
1897 RXAFS_GetVolumeStatus() for each volume ID that is marked vl_busy
1898 or vl_offline. If the volume is now online, the status on the
1899 volume is updated. The default period is 600 seconds. This can
1900 be configured with the BusyVolumeCheckInterval registry value.
1902 * Added prototype for smb_IoctlPrepareRead() which was missing a
1903 return type in the function definition.
1905 * Added volume id lists to the cm_server_t. These lists are
1906 allocated in blocks of ~32 IDs. When a cm_PingServer()
1907 detects a change in server state, the state of the cm_volume_t
1910 * Added volID to the cm_serverRef_t object. volID is used
1911 to identify the volume for which the object is a referral.
1912 cm_FreeServerList() uses the volID to remove the volume
1913 from the cm_server_t.
1915 * In cm_Analyze, when VNOVOL or VMOVED are received,
1916 call cm_ForceVolumeUpdate() to force a refresh of the volume
1919 * Added cm_VolumeStatusNotification() which is used at the moment
1920 to log volume status changes to the trace log. It will also
1921 be used as the access point to the File System Filter driver
1922 notification engine.
1924 * Added an all cm_scache_t list to cm_data. This replaces the use
1925 of the stat cache LRU queue when we need to enumerate all
1926 entries. The LRU list order is not static and when using it to
1927 enumerate all entries it can result in items being missed or
1928 items being processed more than once.
1930 * Modified cm_Analyze(). Instead of reseting the busy or offline
1931 state of a volume and forcing a retry of the operation
1932 cm_Analyze will defer to the background daemon thread that will
1933 update the state once every 600 seconds.
1935 * Added the automatic generation of a Freelance ".root" read-write
1936 mountpoint that refers to the root.afs volume of the workstation
1937 cellname at the time the mountpoint is created.
1939 * re-write cm_Analyze to make better use of the known volume
1940 status. VL_Server queries cannot result in CM_ERROR_ALLOFFLINE
1943 * renamed cm_CheckBusyVolumes to cm_CheckOfflineVolumes.
1944 busy volumes will be reset to srv_non_busy by the function
1945 but there is no mechanism for querying the busy state other
1946 than by attempting to access the resource.
1948 * cm_Analyze will query the state of an offline volume before
1949 deciding whether or not to retry when all volume instances
1952 * the algorithm used after an IP address list change detection
1953 has been modified. we now check the state of the VLDB servers,
1954 then reset the rx connections, and finally check the state of
1955 the file servers. Previously we flushed the rx connections
1956 and then checked the servers in whatever random order they
1957 were first discovered.
1959 * fix the volume status reported by "fs examine"
1961 * various improvements to the AFS Network Identity Manager Provider.
1962 - Don't access freed objects after the window has been destroyed.
1963 - Write back the AFS cell list in the identity configuration
1964 even when its empty.
1965 - Disable all use of the provider when AFS is disabled for a
1966 particular identity.
1968 * Fix cm_cell_t leak. When evaluating a mountpoint that uses an
1969 alias instead of the full cellname, double check the full cell
1970 name and see if we have a matching cell before allocating a new
1973 * Dump more data with "fs memdump". All volumes and scaches and
1976 * Reverse the order of the all cells list. This ensures that
1977 the workstation cell is always first in the list.
1979 * If we recognize the server from which a callback revoke was
1980 received, then we can ensure that we only invalidate the status
1981 information for cm_scache_t objects belonging to that cell.
1983 * When building for amd64, use APPVER 5.02 as we do not support
1984 64-bit Windows 2000 and explicitly state the machine type in
1987 * no longer treat VMWare network adapters as equivalent to a
1988 loopback adapter. For some unknown reason, publishing the "AFS"
1989 netbios name on them frequently results in a name collision
1992 * Use read locks whenever possible instead of write locks when
1993 walking the allCellsp list. This permits better use of multiple
1996 * In 1.5.20 a bug was introduced by the fake status when PRSFS_READ
1997 is not present on the directory optimization. It would report
1998 mount points and directory objects for which there was no
1999 status information as a zero length file. Instead we can use
2000 the vnode number of the object to determine if it should be
2001 considered to be a directory or a file. If it is odd it is a
2002 directory; even it is a file.
2004 * Give Back Callbacks to the file servers in response to suspend
2005 and shutdown power management events. Only give up callbacks
2006 to UP servers and do so with a very short RX dead timeout.
2009 * Fix a race condition in the SMB FID cleanup code that could
2010 result in a stat cache mutex never being released.
2012 * Fix the graphics files that are installed as part of the
2013 IBM Administrators Guide.
2015 * Change the behavior of pioctl path parsing to not follow
2016 symlinks or mount points during evaluation. This results in
2017 the FID being the symlink or mount point and not the object
2018 they point to. This is consistent with the behavior of the
2021 * Enable AFSDB record lookup for command line tools such as
2024 * CellServDB update 2007-07-05
2026 * Added a Volume Status Plug-in Interface that permits third parties
2027 to receive notifications when the volume state changes occur.
2029 * Ensure that all writes to the afsd_alloc.log file use CR-LF.
2031 * Prevent the afsprocmgmt.dll from being unloaded while the
2032 background thread it starts is running.
2034 * Avoid a race condition when closing smb_fid_t objects.
2037 * fix install path of IBM Quick Start Guide when installed by
2040 * fix the client so that it can write into directories with
2041 only 'li' permissions.
2043 * fix installation of 32-bit Shell Extension for 64-bit systems.
2044 Install it under its own HKCR keys.
2046 * modify the wix installer to remove and upgrade install packages
2047 with the same version number.
2049 * when flushing volume location data every two hours or in
2050 response to "fs checkvolumes", also flush the mount point
2051 links in order to force their re-evaluation.
2053 * aklog was copying an uninitialized buffer as part of the
2054 foreign cell automatic pts registration
2056 * when obtaining directory listings, check for read permission
2057 on the directory. If read permission does not exist, fake the
2058 status information for the directory entries. This improves
2059 performance, prevents the client from pounding the file server,
2060 and prevents leakage of cached status information.
2062 * "fs getcacheparms" output was incorrect.
2064 * NetIdMgr AFS Provider should only automatically add token
2065 acquisition for cell foo.com if the realm of the cell matches
2066 the realm of the identity.
2068 * Fix a null pointer dereference in cm_MergeStatus()
2072 * network identity manager provider:
2074 - Correct for the fact that pr_CreateUser tampers with the
2075 contents of the username input buffer.
2077 - Take into account that the krb4 credential data structure
2078 does not have separate realm fields for the user and service.
2079 When krb524 is used with a cross-realm ticket, the realm
2080 describes the service not the user.
2082 * Many integrated logon changes:
2084 - Fix eventlog reporting. Do not attempt to log an event
2085 if the event source registration fails. Use DebugEvent0
2086 instead of DebugEvent when there are no parameters.
2088 - Modify the LOOKUPKEYCHAIN macro to recognize ERROR_MORE_DATA
2091 - Fix the reading of Domain specific configuration for
2092 LogonScript and TheseCells. Previously the dwSize value was
2093 being overwritten so that subsequent RegQueryValueEx call
2096 - Fix a memory leak in the TheseCells reading code.
2098 - Add support for Domain specific "Realm" specification. The
2099 realm is the realm to be appended to the username. When
2100 logging in as a domain or to the local machine, the specified
2101 "Domain" name is not going to be a valid realm name.
2103 - Construct a proper principal name based upon the domain
2104 specified realm for use in obtaining tokens with KFW.
2106 - If the domain specified "TheseCells" list includes the
2107 default cell, do not obtain tokens twice.
2109 * An unprotected call to the Kerberos v5 function,
2110 krb5_free_context(), can result in an invalid memory access
2111 exception if the contents of the krb5 profile, krb5.ini, cannot
2112 be parsed correctly. The error is located within the afskfw
2113 library which is used by afscreds.exe, afssvrmgr.exe, and
2114 afslogon.dll. When the error is triggered within afslogon.dll
2115 a blue screen condition can occur when the winlogon.exe system
2116 process terminates unexpectedly.
2120 * Do not stop the afs client service with an assert
2121 if the cell name cannot be determined when responding
2122 to a "fs whereis" command.
2124 * Integrated Logon (afslogon.dll). No longer attempt
2125 to preserve Kerberos 5 tickets for the logon session
2126 as that job is now supported by KFW 3.1 and above.
2128 * The Network Identity Manager plug-in will attempt to
2129 remove expired service tickets from the cache.
2133 * CellServDB update - GCO Public CellServDB 09 Mar 2007
2135 * AFS Server Manager and AFS Account Manager data cache
2136 moved from HKCU\Software\TransarcCorporation to
2137 HKCU\Software\OpenAFS to prevent crashes due to data
2140 * Reduce the default UP server probe time from 10 minutes
2141 to 5 minutes to match UNIX. This will not solve the NAT
2142 port mapping timeout issue because most NATs default to
2143 a value less that 5 minutes and this value ensures that
2144 the afs client server will start probes 5 minutes after
2145 the last probe set ended. The probe period for any one
2146 server is therefore longer than 5 minutes. The more
2147 servers in the list the longer the time between probes.
2149 * Before marking a server as down in cm_Analyze, retry the
2150 request using a new rx connection to the server. If the
2151 cache manager is behind a NAT, it is possible that the
2152 response from the server is being sent to the original
2153 address:port from which the rx connection was established.
2154 This will appear to the cache manager as the server being
2155 down when in fact it is not. Forcing the use of a new rx
2156 connection will result in the server responding to the
2157 most recent address:port assigned to the client by the NAT.
2159 If the server is in fact down, it will simply take a little
2160 longer to notice it.
2164 * Removing a directory that contained entries was failing with
2165 the wrong error code being returned to the CIFS client.
2169 * rxdebug exponential back-off. don't take down a server that
2170 is already struggling to respond.
2172 * Network Identity Manager plug-in registers help on the Help
2173 menu if it is supported by NIM. (KFW 3.2 and above)
2175 * Plug-n-play networking did not always work after coming back
2176 from a Windows Suspend. The AFS client service failed to
2177 re-compute the LAN adapter to bind to and instead attempted
2178 to bind to all LAN adapters. This would polute the public
2179 network with the name "AFS" which could lead to conflicts
2182 * Build installers using Microsoft Hotfixes provided at
2183 http://tinyurl.com/243svk for Visual Studio C RunTime Library
2184 Daylight Savings Time issues.
2186 * "fs checkserver" does not require admin privileges unless
2187 the probe interval is being changed.
2191 * A reference leak was discovered in the mountpoint traversal
2192 functionality which resulted in cm_volume_t objects not becoming
2193 eligible for re-use. This will trigger a panic condition when
2194 more than the maximum number of volumes is accessed.
2196 * Fixed the success test for the "fs memdump" operation. Success
2197 is now reported as such. Added cm_volume_t object information
2200 * A data corruption error has been fixed. Write requests
2201 queued for background processing were not retried upon failure.
2202 This leads to inconsistencies between the contents of the file
2203 in the local cache and the contents of the file on the file
2206 The implemented fix is to re-queue any request that has failed
2207 due to timeout, volume busy, volume offline, server busy, or
2208 server offline. Requests are not processed while all servers
2209 for a volume are inaccessible. When servers become accessible
2210 the background request will be retried.
2212 In addition, if a background write fails, do not mark the buffers
2213 as clean. Failure to preserve the dirty flag prevents the buffers
2214 from being written to the file server in the future.
2216 This fix does not address the possibility that the credentials
2217 necessary to perform a write request might expire between the
2218 time the request is queued and when the request can be processed
2219 due to server accessibility.
2221 * The Freelance volume does not support per user ACLs. The
2222 cm_HaveAccessRights function did not handle this special case.
2223 Requests for privileges greater than read|lookup would result
2224 in afsd_service.exe entering an infinite loop. ISOBuster is
2225 known to cause this behavior.
2227 * When deleting files, the SMB Server would attempt to obtain
2228 the access permissions for the parent directory instead of
2229 for the file itself. This would result in an incorrect
2230 assessment of whether or not the file can be deleted.
2232 * The Network Identity Manager AFS plug-in has had its string
2233 tables updated for consistency with the language used by NIM.
2234 All credentials are referred to as credentials instead of tokens.
2235 Drop-down combo boxes now in fact drop down.
2237 * The AFS Salvager was not properly being built on Windows.
2238 This is relevant only if the AFS servers are in use.
2240 * The default signal handlers for the AFS servers incorrectly
2241 processed SIGQUIT. Instead of stopping the server, an
2242 exception was raised.
2244 * Fix another volume refcount leak. This one was triggered by
2245 an inability to contact the vlservers in order to update the
2248 * Fix checklist control used by the drive mappings tabs in
2249 afscreds.exe and afs_control.exe.
2251 * Improve output of "fs memdump" related to buffer management.
2253 * Fix refcount leak of cm_buf_t objects when prefetching buffers
2254 already in the process of being fetched.
2256 * Improve cache performance by computing the buffer hash table
2257 size based upon the number of buffers so that the average
2258 number of entries in a bucket is seven.
2260 * Improve cache performance of read/write data by locally
2261 updating the dataVersion of all buffers associated with the
2262 file on each StoreData or StoreStatus operation provided that
2263 the dataVersion was incremented by one and the dataVersion of
2264 the buffer matched the original dataVersion of the cached
2265 status information for the file.
2267 * When recycling cm_scache_t objects, make sure that we
2268 clear all of the previous fields. It is no longer the
2269 case that we can assume that the stat cache object is being
2270 obtained only after a successful FetchStatus call.
2271 InlineBulkStatus will allocate stat cache objects for which
2272 the no status information is available.
2274 * When a file is being opened with the OPEN_ALWAYS disposition,
2275 do not require write permissions when the file already exists
2276 unless of course the file is being opened requesting write
2279 * When a file has been opened for read-only access, if the
2280 application requests an exclusive lock, obtain a read-lock
2281 instead of a write-lock. The purpose of the exclusive lock
2282 in this situation is to prevent changes to the file and the
2283 full file read lock will provide this guarrantee.
2288 * Enforce Unix Mode Write bit to prevent file deletion of read only
2291 * The CIFS READ and LIST permissions use the same value 0x0001.
2292 When a directory is being accessed, check for PRSFS_LOOKUP
2293 instead of PRSFS_READ.
2297 * Fix computations of short names of mount points. They were
2298 broken as part of the directory search optimization. (1.5.13a)
2300 * When all buffers are in use, periodically sleep the threads
2301 waiting for a buffer to become free while the global buffer
2302 lock is not being held in order to permit those threads that
2303 can free buffers to do so.
2305 * Power Management improvements. Maintain a global flag that
2306 specifies whether or not the service is in a suspend state.
2307 Do not panic if Netbios() returns NRC_BRIDGE meaning that the
2308 lana is no longer valid. Instead, stop the listener threads
2309 and if all listener threads are stopped, reset the lana_list.
2310 Allow the cm_Daemon() thread to periodically check the state
2311 of the smb listeners. If they are all stopped and the service
2312 is not suspended, attempt to restart them. If there are no valid
2313 lanas, return to the stopped state.
2315 * CreateX and NTCreateX use cm_CheckNTOpen() to test whether or not
2316 the user is permitted to obtain read or write locks. This function
2317 would obtain the lock and then drop it returning whether or not
2318 the lock could be obtained. If the lock was in fact required,
2319 CreateX/NTCreateX would then obtain it with cm_Lock(). The problem
2320 of course being that this pattern results in three RPCs to the
2321 file server (lock, unlock, lock). This is reduced to one RPC by
2322 implementing cm_CheckNTOpenDone() which frees the allocated byte
2323 range lock from cm_CheckNTOpen() after the cm_Lock() call is
2326 * Optimize the SMB FindFirst processing for the case in which there
2327 are no wildcards. If a cm_Lookup succeeds, generate a response
2328 from the stat cache entry we have a callback for. Its cheaper
2329 than walking each buffer of directory data associated with the
2330 directory stat cache entry when we have thousands of directory
2333 * Increase the SMB Ioctl MaxData size to support larger token sizes
2335 * Disable autorun of the AFS Server Wizard when the AFS Server is
2338 * Add AFS Server service dependencies on tcpip and PNP_TDI
2340 * Remove the AFS Server Configuration Tool shortcut
2342 * Add Freelance mode awareness to AFS Server Configuration Tool
2346 * Unlike previous versions of the OS, Vista performs a shutdown on
2347 the Microsoft Loopback adapter just like it would on a real adapter.
2348 This causes the smb_Listener threads to trigger a panic during a
2349 suspend/hibernate power event.
2351 The fix is to unbind from the network adapters in response to a
2352 suspend/hibernate power event and then rebind to the adapters
2353 when a resume power event is received. Note that the resume events
2354 are not reliably delivered so it is possible the afs service will
2355 not be accessible. However, this is the best we can do.
2357 * don't allow environment variables we set in afslogon.dll to be
2358 inherited by child processes
2360 * force an rx connection reset if VICECONNBAD or VICETOKENDEAD are
2363 * The old dirty buffer synchronization algorithm had a buf_IncrSyncer
2364 thread walking the all buffer list periodically searching for dirty
2365 buffers to write to the file server. This had several negative
2366 results. The algorithm ate up ever increasing amounts of CPU time
2367 even when AFS is idle as the size of the cache increases. Also,
2368 buffers were written to the file server in an order based upon the
2369 original buffer allocation which has nothing to do with the order
2370 in which the buffers became dirty.
2372 The new algorithm maintains a dirty buffer list. Items are added
2373 when the buffer is originally marked dirty. A buffer is only
2374 removed from the list by the buf_IncrSyncer when the buffer is no
2375 longer dirty. If the list is empty the thread goes back to sleep
2376 immediately without additional processing requirements.
2380 * Enable renaming of directories from the Explorer Shell when
2381 Offline Folders is enabled
2383 * An error is reported in response to the SMB ReceiveCoreFlush
2384 request when an error has not occurred.
2388 * optimize dirty buffer flushes in response to suspend operations
2389 do not flush if only loopback adapter is installed
2390 do no retry dirty buffer writes
2392 * do not reset netbios when performing loopback checks. doing
2393 so will break the existing lana bind
2395 * remove a deadlock in smb3.c QPathInfo
2399 * Add NetIDMgr AFS Plug-in
2401 * "Show Tray Icon" checkbox removed from afs_config.exe
2403 * Disable DNS Registrations for the Loopback Adapter and make sure
2404 that Netbios is turned on.
2406 * A reference count leak on stat cache objects was fixed
2408 * For debugging, tread ids of writers are now tracked on locks
2410 * Corrected an incorrect lock that was being held but not
2411 freed in a rare error case
2413 * Changed trace log end of line to CR-LF so it can be read
2416 * modify the buffer synchronization thread to write all dirty
2417 buffers and sleep only when there is no work to be done
2419 * improve behavior of cache manager when large numbers of
2420 temporary files are being created by multiple clients within
2423 * do not permit dirty buffers that belong to deleted files
2424 to be written to the file server. its a waste of bandwidth
2426 * do not leak stat cache references when the file server
2429 * afscreds: do not permit an attempt to obtain tokens if the
2430 afs service could not be started. this was causing stack
2431 overflows on Vista after resuming from a suspend
2435 * better handling of dot directory
2437 * fs sq and fs mkm now behave as on UNIX
2439 * add license to installers
2441 * Return Invalid Parameter error for unknown InfoLevels in
2442 SMB Tran2 Search Directory.
2444 * Correctly use WSAStartup and WSACleanup within the RX library
2446 * Check that afsd_service.exe is running before performing
2449 * Force AFS Shell Extension to load resource DLL using
2450 afs_shl_ext.dll as a base name.
2452 * (Wix) install afs_shl_ext_XXXX.dll and afs_creds_XXXX.dll in
2453 OpenAFS\Client\Program
2455 * Fix SMB_SET_FILE_END_OF_FILE_LENGTH
2457 * Do not panic if the maximum number of volume entries are in use
2458 and one of them can be recycled.
2462 * Fix the code that reads the backconnnectionhostnames value
2463 so that there is not an uninitialized variable
2467 * integrated logon could cause winlogin.exe to blue screen windows
2468 if it was called with a domain that does not exist on the machine.
2469 this is possible during a sysprep.
2471 * corrections to cm_buf_t handling to ensure that buffers are
2472 removed from queues before they are freed and that holds are
2473 obtained on the next buffer in a chain before releasing the
2476 * do not generate MessageBox dialogs on network errors. they
2477 cause more trouble then they are worth.
2479 * move GlobalAutoMapper into a separate thread so that it does
2480 not prevent service startup if the WNetAddConnection2 call
2483 * restrict hard dead and connection timeouts to the CIFS Session
2484 Timeout value. This is being done to try to prevent virtual
2485 circuits from being destroyed prematurely.
2487 * Fix the code that writes the backconnectionhostnames value
2488 to ensure that the data buffer is written with the correct
2491 (there were no windows specific changes between 1.5.4 and 1.5.5)
2495 * Corrections to the Integrated Logon functionality that permits
2496 Kerberos 5 TGTs to be communicated into the user session. The
2497 transfer is now more secure and succeeds when the user does not
2498 have read permission to the %SystemRoot%\TEMP directory.
2500 * NSIS installer updated to version 2.18
2502 * WiX installer updated to version 2.0.4221.0
2504 * Updates to the IFS functionality for 64-bit correctness and
2505 permit compilation. The IFS functionality works for pioctl()
2506 operations but not much else at this point.
2508 * A major re-write of the OpenAFS implementation of the CIFS Tran2
2509 query and set functions.
2511 * Ignore the Windows CIFS client session timeout entirely. Timeout
2512 only after the AFS Hard Dead Timeout period expires.
2516 * move headers, libraries, and samples from OpenAFS\Client to
2519 * remove the auto-registration of AFS IDs for foreign cells
2520 from integrated logon. this permits afslogon.dll to avoid
2521 initializing the RX library. afslogon.dll is a network
2522 provider library and as such it is loaded into all applications
2523 which perform a CreateProcessAsUser call. Initializing the
2524 RX library spawns background threads to listen for inbound
2525 packets. When the network provider is unloaded from the
2526 process, the RX library does not cleanup the threads. If they
2527 continue to execute after the library is unloaded, the process
2530 * implement inline bulk status calls. this will significantly
2531 improve the performance of directory listings.
2533 * prevent a crash when evaluating mountpoints to volumes that do not
2536 * new command: fs uuid [-generate]
2537 displays the current uuid and allows a new one to be generated
2538 on the fly without restarting the afs client service.
2540 * fix the 64-bit installers to properly document the install
2543 * move help files to the proper location so they can be located by
2544 the system tray tool and the control panel.
2546 * "over quota" and "disk full" errors were not being reported to the
2549 * largefile support has been added. Files greater than 31-bits can no
2550 longer be written to file servers which do not support large files.
2551 When file servers support large files, files up to 64-bits can be
2554 * cm_SyncOp could return with the input buffer unlocked upon error
2555 which would result in a panic condition. This bug was introduced
2558 * vlserver failover when freelance mode did not work. This prevented
2559 the root.afs volume from being loaded if the first vlserver could
2564 * A bug in the CIFS Standard Info response packet caused directories
2565 to appear as standard files. This prevented Microsoft Vista from
2566 being able to walk the AFS directory tree.
2568 * Mount points must be reported as directories in response to
2569 Query Path Info operations.
2571 * A deadlock was detected when performing "fs flushall" if the
2572 file server reports VNOVNODE. The scp->createBufferLock is
2573 already held by the current thread and there is no mechanism
2574 to propagate the knowledge. Therefore, an alternate mechanism
2575 for clearing the cache must be developed. A new function
2576 cm_RecycleSCache(scp,flags) has been extracted from
2577 cm_GetNewSCache(). This function performs the task of recycling
2578 an cm_scache_t object. When called from cm_GetNewSCache() with
2579 no flags the expectation is that there are no associated buffers
2580 that are queued to be read or written. When called from
2581 cm_Analyze() with the CM_SCACHE_RECYCLEFLAG_DESTROY_BUFFERS
2582 flag, any queued buffers will be de-queued and marked as if
2583 the operations were performed so the data can be discarded.
2585 * CM_ERROR codes were logged as "unknown" by cm_Analyze.
2587 * In response to VNOVNODE, the parent is only discarded if the current
2588 cm_scache_t is not a directory.
2590 * In the Ioctl FlushFile and FlushVolume functions, there were no
2591 checks to protect against flushing the Freelance SCache entries.
2593 * In FlushFile, the wrong cm_scache_t object was being released.
2595 * In cm_GetNewSCache, do not allow Freelance SCache entries to be
2596 recycled. Choose a new entry if cm_RecycleSCache fails.
2598 * Speed up the performance of the cache manager by not
2599 holding the smb_fid_t mutex across calls to cm_SyncOp
2602 * Ensure that all smb_fid_t flag references are protected.
2604 * Remove a deadlock between smb_fid_t mutex and
2605 smb_rctLock introduced in 1.4.1.
2607 * Fix aklog AFSID lookup to use not send the realm
2608 name when the realm is the local realm for the
2609 cell. The PT Server doesn't strip the realm
2610 name when it is the local realm.
2612 * Treat "\\afs\*." as an alias for "\\afs\all"
2614 * Dynamically adjust the priority of server threads
2615 based upon the age of the CIFS request that is
2616 being processed. The priority is bumped one level
2617 for every 15 seconds of processing time.
2619 * The original openafs contribution mis-used the
2620 cm_GetCallback function. This function should only
2621 be called as a side effect of a call to
2622 cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK). The way it
2623 was being called results in a FetchStatus RPC being
2624 performed even when the client already has a valid
2625 callback and does so in a manner that destroys the
2626 synchronization of the threads that are calling
2627 cm_SyncOp. If nothing else this results in a
2628 significant performance penalty.
2630 * More changes to cleanup of smb_vc_t objects in
2631 order to prevent race conditions.
2633 * Reduce the amount of Power Management event logging
2636 * If the client detects that the IP address has
2637 changed, force the use of new RX connections. This
2638 enables the file server to recognize that the client
2641 * Remove a race condition and deadlock associated
2642 with cleanup of smb_vc_t objects.
2644 * Fix the ANSI filename support. One function that
2645 required it was missing the translations.
2647 Changes not present in 1.4.1 and since 1.5.0: 1.5.1 released
2650 * Add a DllMain() function to afspthread.dll. Perform
2651 cleanup of thread local storage, rx queues, critical
2652 sections, event handles, etc. This was implemented
2653 to avoid crashes in afspthread.dll when applications
2654 which load/unload the dll attempt to terminate.
2656 * Lock request behavior has been altered. As of this
2657 release a locally allocated lock will be used whenever
2658 the volume is RO or the requesting user's permissions
2659 are 'rl'. This will allow end users to execute programs
2660 from RW volumes when they do not have the 'k' privilege.
2662 Note that a recent discovery indicates that the 'k' bit
2663 is only supposed to control the ability to obtain read
2664 locks. Write locks are to be controlled by the 'w' and
2667 * Code has been added to enable the cache manager to be
2668 able to obtain the AFSID of the authenticated user.
2669 This is required to be able to determine when "Insert"
2670 privileges should apply. AFS considers a file to be
2671 a new file whenever the owner's AFSID matches the
2672 AFSID of the user. The current implementation relies
2673 on the name specified during the SetToken call to be
2674 correct. This will be replaced in the future with a
2675 new RPC that allows the AFSID of the authenticated
2676 user to be obtained.
2678 * A 32-bit-tools installer for use on 64-bit systems
2679 is now available. These tools are for use with 32-bit
2680 versions of Kerberos for Windows or JPSoftware's 4NT.
2682 Since 1.4.1 rc10: 1.4.1 final released (9 April 2006)
2684 * Remove the use of the "active_vcp" variable. This variable
2685 no longer serves any useful purpose and was the cause of
2686 race conditions that produced under counts of the smb_vc_t
2689 * Change default minidump output to include data segments
2691 * restore access to cm_IoctlGetVolumeStatus and
2692 cm_IoctlSetVolumeStatus calls for AFS pioctl operations.
2693 Still need to figure out why they are being called
2694 for SMB/CIFS IOCTL operations instead of Trans2GetFileInfo.
2696 Changes not present in 1.4.1: 1.5.0 released (16 Feb 2006)
2698 * Update build system to include the "sys_name" in the
2701 * optimize the number of calls to obtain the smb_user_t for a
2702 smb virtual circuit.
2704 * add logging of server addresses, status and capabilities when
2707 * prevent multiple threads from pinging the same server simultaneously.
2708 instead, threads block and then use the result of the ping that was
2709 already taking place.
2711 * Add registry values to allow the defaults for daemon checks to be altered
2712 at startup. Log values to afsd_init.log.
2714 daemonCheckDownInterval
2715 daemonCheckUpInterval
2716 daemonCheckVolInterval
2717 daemonCheckCBInterval
2718 daemonCheckLockInterval
2719 daemonCheckTokenInterval
2721 * Add EventLog messages for transitions in fs crypt state.
2723 * Byte range locking (when locks will be obtained from the
2724 is still under discussion)
2726 * 64-bit Windows operating systems now supported
2727 (XP SP2 64-bit, 2003 SP1 64-bit, 2003 R2 64-bit)
2728 This is a native 64-bit implementation of OpenAFS for Windows.
2729 There is currently no 64-bit version of KFW and therefore
2730 there is no (easy) method for obtaining AFS credentials.
2731 32-bit KFW 3.0 can be used with 32-bit OAFW libraries
2732 to obtain AFS tokens but there is not for the feint of
2735 * A faster implementation of the fcrypt algorithms improves
2738 * The afsd_service.exe now supplies Event Log Message Text.
2739 The event log no longer generates warning text due to lack
2740 of message configuration.
2742 * When afs servers are unreachable and network stack returns
2743 a WSAEHOSTUNREACHABLE error, the AFS client will immediately
2744 mark the attempted server as being down. This allows for
2745 very fast response to the calling application. This change
2746 has a pronounced improvement when the Windows Explorer is
2747 used on disconnected mobile computers.
2749 * Added DebugOutputString debugging of the RX library
2751 Since 1.4.1 rc9: 1.4.1 rc10 released (10 March 2006)
2753 * Fix the rx-lwp implementation to randomly generate rx call
2754 identifiers which are supposed to be unique at a given point
2755 in time. This affects some of the command line tools used
2756 for debugging but not the AFS Client Service itself which
2757 uses the rx-pthreads implementation.
2759 * xstats_fs_test.exe and xstats_cm_test.exe are now built for
2760 Windows although they are not distributed
2762 * don't generate an exception when logging unexpected SMB packets
2764 * the data structures used by cm_IoctlGetVolumeStatus and
2765 cm_IoctlSetVolumeStatus are not compatible with the CIFS
2766 Get/Set File Information calls so don't bother executing them.
2770 * Fix data written to registry as part of BackConnectionHostnames
2773 * Fix initialization bug when allocating cm_server_t objects that
2774 would cause servers to appear to be down after each allocation
2775 or renewal. Server would appear up after the next check servers
2780 * Windows will return WSAECONNRESET to the next receive packet
2781 operation on a socket if an ICMP packet is received in response
2782 to a packet that was sent and could not be delivered because
2783 the port was unreachable. We have to treat this as
2784 WSAEWOULDBLOCK so that we can retry the select() check and
2785 not mark the connection as bad.
2787 Since 1.4.1 rc6: 1.4.1 rc7 released (14 Feb 2006)
2789 * The client manager maintains a least recently used list of
2790 all the ACLs. If an ACL is already the most recently used
2791 do not bother to remove it from the list and place it on the
2794 * As part of the process of adding a new server, set the
2795 server is down flag and then perform a ping. If the server
2796 is up, the flag will be reset. If the server is down, the
2797 client will not timeout unnecessarily while attempting to
2800 * If a Mountpoint String is the empty string, return PATH NOT
2803 * Modify behavior of the Network Provider to avoid querying the
2804 profile location if integrated logon is known to be disabled.
2806 * Correct behavior of 'LogoffPreserveTokens'. When set, tokens
2807 will always be preserved regardless of whether or not integrated
2808 logon was used to obtain tokens. When unset, tokens will be
2809 destroyed regardless of whether integrated logon was used to
2810 obtain the tokens unless the profile is stored in AFS.
2812 * Disable LogoffTransferToken functionality. The SMB V3 Logoff
2813 message is being delivered even when the user is not logging
2814 off of the computer.
2816 * Fix a deadlock condition introduced in rc6.
2818 * Cell names are case-insensitive. Since the founding of OpenAFS
2819 there has been a bug that treated "cell", "Cell", and "CELL" as
2820 different cells. This would result in what appeared to be
2821 random disappearance of user tokens or a failure to authenticate
2822 even when tokens were listed by the "tokens" command.
2824 Since 1.4.1 rc5: 1.4.1 rc6 released (2 Feb 2006)
2826 * Additional auditing of the smb code revealed that smb ioctl
2827 operations were failing to free smb_fid_t objects
2829 * The File Identifiers assigned to smb_fid_t objects could be
2830 given the value 65535 which is 0xFFFF which is also the
2831 INVALID_HANDLE value. Both 0 and 65535 are now treated as
2834 * Audited all assignments to fields and ensured that they are
2835 protected by obtaining the relevant mutex object.
2837 * Re-wrote the VNOVNODE handling to avoid a potential deadlock
2838 if both the current cm_scache_t object and its parent are
2839 both returning VNOVNODE.
2841 * Removed all the logic associated with the 'dead_vcp' variable
2842 which must have been added for debugging because its used
2843 was never thread safe and it truly served no purpose.
2845 Since 1.4.1 rc4: 1.4.1 rc5 released (25 Jan 2006)
2847 * Re-enabled LogoffTokenTransfer and LogoffTokenTransferTimeout
2848 registry settings. Default is ON and 120 seconds. If set to
2849 OFF, tokens will not be destroyed.
2851 * Fixed a number of race conditions and reference count problems
2852 associated with global queues.
2854 * Added a method to detect abandoned smb virtual circuits so their
2855 associated file handles, locks, user tokens, and smb session
2856 resources can be released.
2858 * Added a mechanism to allow user tokens to be preserved
2859 during integrated login and be freed after logoff.
2861 * Check access permissions to the file before the file is
2864 * More improvements to the handling of SMB Virtual Circuits
2865 when SMB sessions are closed.
2867 * Correct a race condition introduced in rc4 that could result
2868 in a crash of afsd_service.exe.
2870 Since 1.4.1 rc3: 1.4.1 rc4 released (11 Jan 2006)
2872 * Properly handle pre-mature SMB session closures.
2873 File handles, tree connections, and user sessions are now
2874 freed when the virtual circuit is closed.
2876 * smb_username_t and smb_user_t objects are now properly reference
2877 counted and freed as appropriate.
2879 * disable to use of AFS file locks. byte range locking is still
2880 implemented but all locks are managed locally. This prevents
2881 multiple processes from editing the same file on the same
2882 machine but it does prevent multiple processes on separate
2883 machines from modifying the file.
2885 Since 1.4.1 rc2: 1.4.1 RC3 released (4 Jan 2006)
2887 * A fix to "fs setserverprefs -vlserver". Multiple calls with
2888 the same server parameter would result in a panic condition
2889 if the server had not already been contacted by the AFS
2892 * Better warnings are written to the afsd.log file when
2893 the Windows RPC Protocols are not properly configured as
2894 is often the case on Windows 2000.
2896 * fix the procmgmt library to restore C RTL signal handlers when
2897 it is being detached from the process prior to process termination.
2899 Since 1.4.1 rc1: 1.4.1 RC2 released (30 Nov 2005)
2901 * close a lock leak in the rx library that could be triggered while
2904 * prevent removal of drive mappings in response to afscreds.exe -M
2906 * add logic to process VNOVNODE in cm_Analyze. Force re-evaluation of symlink
2907 strings and flush the stat cache entry.
2909 * prevent server objects from being freed if user preferences are set.
2911 * Update WIX support to 2.0.2419.0
2913 * Release all locks on the server at service shutdown.
2915 * A down server should be marked up as long as it responds to the client.
2916 It doesn't matter if the server returned an error as long as the error
2917 is something other than VRESTARTING. If there is such an error, all
2918 connections on that server should be marked for replacement.
2920 * Change the check "up" servers period from once per hour to once every
2921 ten minutes. This matches the behavior of the Unix cache manager.
2923 Post 1.4.0 revisions: 1.4.1 RC1 released (21 Oct 2005)
2924 * Byte Range Locking support has been added. Architectural documentation
2925 is located within src/WINNT/afsd/cm_vnodeops.c. Best efforts are
2926 made to enforce Windows mandatory locking model using locally managed
2927 byte range locks with full file locks obtained from the file server.
2928 Read-only volumes provide the user with an implied locking privilege.
2929 Read-write volumes require that a user be granted the locking privilege
2930 in order for a lock to be obtained. As most Windows applications use
2931 Shared file access modes, they require the ability to obtain locks in
2932 order to access files.
2934 * Implement error translation functions to convert Universal Error Codes
2935 output by modern servers to local errno and winsock error values.
2936 This fixes the error message problems using pioctl() calls from fs.exe
2937 and the AFS Shell Extension.
2939 Since 1.4.0 rc8: 1.4.0 final tagged (20 Oct 2005); released (31 Oct 2005)
2940 * Allow the AFS Client Service to set the firewall configuration by
2941 retrying until after Windows determines the boot period is over.
2943 * Do not attempt to configure the firewall on Windows 2000, XP pre-SP2,
2946 * Detect IP address changes from within the AFS Client Service for the
2947 purpose of checking down servers sooner
2949 * If the file server or vldb server report RXKADUNKNOWNKEY, return
2950 SEC_E_NO_KERB_KEY to Windows. This provides an indication to the
2951 user that there is a Kerberos key problem. Previously the user would
2952 receive no feedback when there was something wrong.
2954 * Fix the large integer library support to properly handle values greater
2957 Since 1.4.0 rc7: 1.4.0 rc8 released (15 Oct 2005)
2958 * Updated CellServDB
2960 * Fixed file name pattern matching. Question marks can match nothing
2961 if found at the end of a component.
2963 Since 1.4.0 rc6: 1.4.0 rc7 released (8 Oct 2005)
2964 * Kerberos 5 ticket lifetimes should adhere to KFW Leash lifetimes
2966 * The MSI installer now has a new Product ID for each build. This
2967 is to allow for automated upgrades. The Upgrade ID is remaining
2970 * Volumes that were cached and then subsequently were salvaged,
2971 moved or otherwise taken temporarily offline (as opposed to being
2972 busy) would become unreachable. A distinction is now made between
2973 a volume being offline and a server being down.
2975 Since 1.4.0 rc5: 1.4.0 rc6 released (2 Oct 2005)
2977 * Fixed several AFS Shell Extension issues:
2978 - Checking Server Status would only work if a specific cell was
2980 - If servers were down, the Server Down dialog would write to
2981 invalid memory and cause explorer.exe to crash
2982 - Default Push Buttons were not specified for many dialogs
2983 - The extension would be unloaded by the Explorer Shell and
2984 never be reloaded under some circumstances.
2986 * Fixed the AFS System Tray menu to disappear if another Window
2987 was selected before a menu item was selected
2989 * Added volume owner and group information to "fs examine" output
2991 * Added fs and registry support for enabling/disabling/clearing
2992 rx statistics gathering.
2994 Since 1.4.0 rc4: 1.4.0 rc5 released (25 Sep 2005)
2996 * New Product Code GUIDs issues for 1.4 and 1.5 branches. Separate
2997 GUIDs for each platform.
2999 * Documentation directory structure and default installation rules
3002 * New OpenAFS for Windows HTML Release Notes installed
3003 as part of the documentation set.
3005 * Add thread id to rx debugging messages
3007 * When all servers are offline, return a bad network path error
3008 immediately. This is necessary to allow Explorer to do the
3009 right thing and not hang. The background daemon thread will
3010 check the down servers every 30 seconds to see if they are
3013 Since 1.4.0 rc3: 1.4.0 rc4 released (14 Sep 2005)
3015 * The AFSCache file has been modified to store both the
3016 serial number of the volume on which the file resides
3017 and the SID of the local machine. If either change,
3018 a new UUID is generated.
3020 * add "fs flushall" command that forces all data buffers
3023 Since 1.4.0 rc2: 1.4.0 rc3 released (4 Sep 2005)
3025 * add support for '/' instead of '\' in pioctl() calls
3027 * Apply AFS Client Admins group protect to AFS Shell Extension
3029 * Add support for \\afs\<name> to most AFS Shell Extension
3030 commands except the symlink methods.
3032 * when installing the Microsoft Loopback Adapter, enable
3033 MS Client for Networks and prevent an install failure
3034 by not calling CoInitialize twice in the same thread.
3036 * reload cell vldb values from the CellServDB every two
3037 hours in case it changes
3039 * When updating cell information from DNS, be sure to set
3042 * Add support to allow use of \\AFS\<foo> where <foo> is
3043 either a mount point or symlink. As <foo> is normally
3044 treated as a share name, we transform it into \\AFS\all\<foo>
3047 * Init 'code' to prevent false errors when integrated logon
3048 is disabled and the service is not running
3050 Since 1.4.0 rc1: 1.4.0 rc2 released (28 Aug 2005)
3052 * Removed trace log messages that were hampering performance.
3054 * Fixed a deadlock that was being triggered by editing Word
3055 documents stored within AFS with WinWord 2003.
3057 * Bit 3 of the TraceOption registry value is now used to set the
3058 default for "fs trace" in the non-Debug builds. The new default
3059 is off for release builds and on for debug builds.
3061 * Bit 2 of the TraceOption registry value can be used to turn on
3062 real-time output of debug log entries to the Windows Debug
3063 Output monitor interface. This data can be viewed with tools
3064 such as Sysinternal's DbgView.
3066 Since 1.3.87: 1.4.0 rc1 released (17 Aug 2005)
3067 * When the cache manager reports ALLBUSY or ALLOFFLINE for
3068 a resource we should not be returning status codes such
3069 as Network Busy or Remote Host Down because these errors
3070 will imply that the AFS Client Service is Busy or Down
3071 and that is simply not the case. Instead we will return
3072 Bad Network Path as the path is temporarily not available.
3074 Instead of returning Sharing Paused when there is a
3075 timeout error, return Timeout. Once again, the AFS Client
3076 Service is not paused.
3078 * afscreds.exe would display an Obtain Creds dialog when
3079 the expired credentials reminder was triggered even
3080 if there was no network path to the KDC. This is prevented
3081 by adding KDC probe logic to the reminder thread.
3083 * afscreds.exe would display expired tokens no differently
3084 than unexpired ones. This would make it difficult for the
3085 user to distiguish when the tokens were expired. For the
3086 English build added a new resource string "(expired) that
3087 is displayed instead of the expiration time.
3090 * "fs wscell" when executed in freelance mode will return the
3091 name of the cell configured in the registry. The root.afs
3092 volume is not loaded from this cell, but it is used for the
3093 default for aklog and integrated login.
3095 * "fs mkmount, fs rmmount, symlink make, symlink remove" will
3096 no longer work on \\AFS\all when freelance is being used
3097 unless the user is a member of the "AFS Client Admins" Windows
3100 * some more checks performed during persistent cache validation.
3103 * Optimize calls to cm_CheckServer from cm_Analyze. Only check
3104 down servers for the current cell not all cells.
3106 * If the scache LRU list becomes corrupted, the AFS Client Service
3107 might enter an infinite loop at startup. Detect the corruption
3108 and discard the cache contents in this case.
3110 * Fixed the Explorer Shell's remove mountpoint functionality.
3112 * Fixed a deadlock caused by the holder of cm_aclLock attempting to
3113 obtain a mutex lock on a cm_scache_t object whose lock is held by
3114 a thread attempting to obtain the cm_aclLock.
3117 * Really fixed DNS AFSDB queries to ensure that "csail" cannot be
3118 misinterpretted as "csail.mit.edu" when the resolver
3119 is configured to append ".mit.edu" to failed DNS queries.
3121 * Added a new registry key, "LogoffPreserveTokens" (see registry.txt),
3122 that can be used to force the preservation of user tokens upon logout.
3124 * Update the NSIS install scripts to use NSIS 2.07. This release adds
3125 recommended installation categories: AFS Client, AFS Administrator,
3126 AFS Server, AFS Development Kit. Each category includes a different
3127 default subset of the OpenAFS components.
3129 The OpenAFS logo is now associated with the NSIS uninstall entry
3130 in the Add/Remove Programs control panel.
3132 * The user name associated with AFS tokens when obtained with
3133 integrated login, the afs systray tool, or aklog will always
3134 include the full Kerberos 5 user principal name regardless of
3135 whether or not the cell is local to the realm.
3137 * Modify integrated login so that it does not enter an infinite
3138 loop if the service is not set to auto start.
3140 * Added asetkey.exe used to set a Kerberos 5 key for use by the
3143 * Added uninstall.exe to Wix installation
3145 * More modifications to algorithms used to wake sleeping threads.
3149 * Changes to the algorithms used to wake threads when they are
3150 put to sleep because of conflicts over buffer operations.
3153 * The OpenAFS integrated logon obtains Kerberos 5 tickets as part of
3154 the process of producing AFS tokens. The tokens are stored within
3155 the AFS Client Service but the Kerberos 5 tickets have been discarded.
3156 New functionality has been added to temporarily copy the tickets to
3157 a file ccache during the short window between the NPLogonNotify and
3158 WinLogon Logon Event. The Logon Event handler starts a process as
3159 user, afscpcc.exe, that copies the tickets into the default ccache.
3160 The file ccache is then destroyed by the event handler.
3162 * During synchronization operations on stat cache entries, if threads
3163 are forced to sleep it was possible that the threads would not be
3164 woken under heavy load.
3166 * Enforce the delete on close flag specified by the NTCreateX()
3169 * A race condition in the rx library was fixed that resulted in
3170 a deadlock in rx_NewCall when the RX_CONN_MAKECALL_WAITING flag
3171 was cleared when threads were still waiting to use the
3172 connection. Also, fixed a potential case in which the
3173 connection could be destroyed while threads were waiting to use it.
3175 * The DNLC freelist has been seen to become corrupted with still
3176 active entries being stored on the freelist. Changes were made
3177 to perform a better job of cleaning entries before placing them
3178 onto the freelist; marking them with the correct magic value when
3179 purging the list; and allowing the list to be purged on startup
3180 if the cache validation fails.
3182 * Windows Crash Reporting does not get triggered for afsd_service.exe
3183 because it provides its own top level exception handler. This patch
3184 places an explicit call to ReportCrash() on platforms which support
3187 If you configure Crash Reporting via AD Policy to capture crash
3188 reports within the domain, then you will receive this output.
3190 * Generate MiniDumps independent of the Windows Crash Reporting.
3192 (a) If an exception occurs,a minidump will be generated at the path
3193 %TEMP%\afsd.dmp. The type of minidump defaults to Normal but
3194 can be set to any other type via use of the
3197 [HKLM\SOFTWARE\OpenAFS\Client] MiniDumpType (DWORD)
3199 (b) MiniDumps can be generated at any time using the "fs minidump"
3200 command. This allows you to generate a minidump without
3201 stopping the process or even requiring the use of a debugger
3204 * an RX connection which reports bad ticket now treats it the same way
3205 the expired ticket is treated. The ticket is removed from the
3206 connection and a new anonymous connection is established.
3208 * fs memdump now requires that the caller be in the AFS Client Admins
3211 * additional debugging information is provided in cm_Analyze when the
3212 error code is ignored.
3214 * Fix a regression introduced into the pioctl() client code which
3215 will cause a crash in winlogon.exe, explorer.exe and fs.exe if
3216 MIT KFW is not installed. (1.3.82a)
3218 * Remove AFS Gateway option from OpenAFS Control Panel (1.3.82a)
3221 * Fix a race condition in cm_EndCallbackGrantingCall() that
3222 could leave 'nrevp' pointing to freed memory. In the process,
3223 optimize the processing to ensure that the applications monitoring
3224 the status of the stat cache entry are only notified at most
3227 * Conditionalize access to the Explorer Shell AFS->Mount Point
3228 ->Remove menu item. Disable the item if the selected item(s)
3229 are not mount points.
3231 * Activate AFS RPC (RX) Free Packet Queues stored in Thread
3232 Local Storage. This eliminates thread contention on the
3233 global RX free packet queue. This should improve performance
3234 on multi-CPU systems.
3236 * Fix the IP address filtering code to properly load/unload
3237 the IPHelper DLL so that Windows XP/2003 and beyond systems
3238 do not have to manually probe the registry.
3240 * CleanupACLEnt() was not being called consistently with
3241 the cm_scache_t object referred to by the back pointer
3242 mutex locked. This could in very rare conditions lead to
3243 an invalid memory access.
3245 * Added a script command to the msi installer "afs_replace"
3246 which can be used to replace one version of OpenAFS with
3247 another without requiring an uninstall and reinstall if
3248 the installed msi is no longer in its original location.
3250 * Apparently the problem with multi-domain forests with cross-
3251 realm trusts to non-Windows realms was not entirely solved.
3252 The authentication to the AFS SMB service failed because
3253 the wrong name was being used. Using ASU as an example,
3254 the authentication was being performed with the name
3255 "QAAD\user" (an account in the forest root) and not
3256 "user@ASU.EDU (the MIT Kerberos principal used to login with)
3258 The solution was to add an additional dependency on KFW
3259 in order or to be able to easily obtain the client principal
3260 name stored in the MSLSA ccache TGT. This information is
3261 used in two locations:
3263 - the pioctl() function
3265 - a new WinLogon Event Handler for the "logon" event.
3267 The pioctl function will now be able to use the correct
3268 name when calling WNetAddConnection2() and the "logon"
3269 event handler will now be able to call WNetAddConnection2().
3270 The hope is that the "logon" event handler will be called
3271 before the profile is loaded but I have not guarrantee
3275 * Fixed a locking error in cm_TryBulkStat() which had the
3276 potential of deadlocking the system for the length of time
3277 it takes to perform a bulk status fetch operation.
3279 * Replaced time conversion code (UnixTime <-> FILETIME) to
3280 be completely arithmetic instead of relying on a bizarre
3281 algorithm involving a variety of C RTL time functions.
3282 This has the side effect that UnixTime and FILETIME which
3283 are both stored in UTC are interpretted as UTC throughout
3284 the year. Windows will apply the same localization to AFS
3285 as it does to NTFS. Applications which rely on the ability
3286 to sync files between the two file systems will no longer
3287 see the timestamps of files in AFS change an hour relative
3288 to the files stored in NTFS or Windows based backup devices.
3290 * Fixed a invalid memory access under a bizarre circumstance.
3291 Windows will allow a physical mass media device to be
3292 installed via Plug N Play to the system and will assign it
3293 the lowest available drive letter. This is true even if the
3294 drive letter is currently assigned in the user session to
3295 a network device via NET USE (or its equivalent.) When this
3296 happens, queries sent to the CIFS server will contain invalid
3297 data. This invalid data was not being caught by the AFS
3298 Client Service and was resulting in a crash when Freelance
3301 * Fixed a reference count error when registering callbacks
3302 on a stat cache object if the callback was already registered
3304 * Add a case to cm_Analyze() in order to handle the case where
3305 cm_GetConnByMServers() returns CM_ERROR_NOSUCHVOLUME because
3306 the server list for the volume is empty. In this case, force
3307 an update of the volume info and retry.
3309 * Insert a missing cm_EndCallbackGrantingCall() which could
3310 result in threads waiting for a callback to complete to never
3313 * In the persistent cache, there is a maximum number of volume
3314 entries. Allow volume entries to be reused if the maximum
3315 number have been allocated and their reference count is zero.
3317 * If we already have a dead virtual connection object, don't
3318 ignore the fact that additional dead vc objects must be taken
3321 * Removed a deadlock condition introduced in the summer of 2004.
3322 Do not hold mx locks around calls to RX functions including
3323 rx_NewCall(). That is what reference counts on the rx_connection
3326 * Fixed an initialization error in afslogon.dll which could
3327 result in random behavior including a failure to terminate
3328 the NPLogon function. (1.3.80b)
3330 * Fixed an error preventing the use of SMB authentication
3331 on Windows 2000. (1.3.80a)
3334 * Updated CellServDB to Public CellServDB 16 Mar 2005
3336 * Fixed DNS AFSDB queries to ensure that "csail" cannot be
3337 misinterpretted as "csail.mit.edu" when the resolver
3338 is configured to append ".mit.edu" to failed DNS queries.
3340 * Fixed another case in which the client would replace connections
3341 to the host when it was not necessary. In this case the
3342 situation would occur if cryptall was on and the connections
3343 were unauthenticated due to lack of an appropriate token.
3345 * OpenAFS for Windows has failed to work at sites which are
3346 utilizing a cross-realm trust between an MIT/Heimdal realm
3347 and a multi-domain Windows forest when the workstation being
3348 accessed is not located in the root domain. This is caused
3349 by a bug in the workstation which was triggered after the
3350 introduction of Windows 2003 Server. When the bug is triggered,
3351 the workstation attempts to authenticate users to afsd_service.exe
3352 by contacting the Domain Controller instead of using the
3353 LSA loopback authentication mechanism.
3355 One of the reasons this bug occurs is because the workstation
3356 does not have a reliable way of knowing that the service whose
3357 netbios name is "AFS" is located on the workstation. This will
3358 be fixed starting in Longhorn Beta 1 by Microsoft. The
3359 "BackConnectionHostNames" registry value will be used to
3360 indicate that the authentications to that service name should
3361 be performed using the loopback authentication mechanism.
3363 In the meantime, when Logon Caching is enabled, we can force
3364 afsd_service.exe to authenticate using the logon cache before
3365 contacting the Domain Controller. This will work with both
3366 password and smart card based logons.
3368 * The allDown logic in cm_ConnByMServers() was wrong. The allDown
3369 flag should not be cleared if a volume's server reference is
3370 marked as "offline". In the case where all of the volume's
3371 servers are either "down" or the volumes are "offline", we want
3372 cm_Analyze() to process the condition as CM_ERROR_ALLOFFLINE
3373 instead of as CM_ERROR_TIMEDOUT. In fact, CM_ERROR_TIMEDOUT
3374 should never occur in practice.
3376 In the case of CM_ERROR_ALLOFFLINE, cm_Analyze() will sleep for
3377 5 seconds, clear the server down and volume busy flags, and
3378 then force an update from the VLDB. This allows the client to
3379 update the location of a volume if the reason for it being
3380 marked offline is because it is being moved. Calls to
3381 cm_ConnByMServers() will be retried either until success or
3382 the RDRtimeout period is reached.
3384 * Correct the Power Management code behavior in response to
3385 Standby, Suspend, and Shutdown events. Instead of flushing the
3386 buffers associated with the mounted SMB submounts, simply write
3387 all buffers which are marked dirty.
3389 * Added support for "TheseCells" to afslogon.dll. "TheseCells"
3390 provides a list of cells other than the default cell for which
3391 tokens should be obtained using the default Kerberos principal.
3392 This functionality is logon domain specific and is only available
3393 when using KFW for authentication.
3395 * Fixed FindNext Invalid Handle error caused by over agressive
3396 attempts at garbage collecting dirSearch entries when the
3397 dirSearch ID wraps from the maximum value to 1.
3399 * Add support for registry defined server preferences for VLDB and
3400 File servers. See registry.txt for details.
3402 * Increased default cache size to 96MB and default number of cache
3405 * Fixed refCount leaks related to directory lookups and pioctl calls.
3407 * Callbacks revoked during a race condition with an attempt to
3408 obtain the same callback no longer result in an inaccessible
3411 * New command line tool:
3413 afsdacl : Set or reset the DACL to allow starting or stopping
3414 the afsd service by any ordinary user.
3416 Usage : afsdacl [-set | -reset] [-show]
3417 -set : Sets the DACL
3418 -reset : Reset the DACL
3419 -show : Show current DACL (SDSF)
3421 * IP addresses are no longer queried once at startup. Instead IP
3422 addresses are obtained as needed. Loopback adapter addresses are
3423 no longer published to the server.
3425 * Pay attention to the MIT KFW registry configuration for automatic
3426 importation of MSLSA credentials. (SOFTWARE\MIT\Leash32,MsLsaImport)
3428 * Fix (once again) case-sensitive comparisons which was apparently
3431 * Activate support for vos listvol -format
3433 * Population of the "AFS Client Admins" group failed on non-English
3434 versions of Windows because the name of the "Administrators" group
3435 is localized. Now we lookup the name of the group by using the
3436 Administrators Group Alias Relative ID.
3438 * Multiple cell token acquisition within afscreds.exe was broken in the
3439 case of Kerberos 5 cross realm authentication. Instead of contacting
3440 the KDC associated with the cell's realm, afscreds would obtain a
3441 token from the KDC of the user's realm. This would result in
3442 "invalid kvno errors" while authenticating to the AFS servers.
3443 Unauthenticated access would work. The symptoms would vary based upon
3444 whether or not the VLDB servers had been contacted using unauthenticated
3445 connections prior to the user obtaining tokens.
3447 * The list of ACL entries was becoming corrupted. It appears as if
3448 an ACL when it expires was not being appropriately cleaned up. In
3449 fact, it was left in the list of ACLs associated with the scache
3452 * Changed the default @sys name list to "x86_win32 i386_w2k i386_nt40"
3453 for 32-bit x86 systems. The default for itanium will be "ia64_win64"
3454 and "amd64_win64" for amd 64-bit processors.
3456 * When executing executables, Windows provides a mechanisms by which
3457 the normal search for DLLs can be bypassed. If a file foo.exe is
3458 being executed and there exists a file or directory "foo.exe.local"
3459 (the contents are ignored if a file), then Windows will search for
3460 DLLs first in the "foo.exe.local" directory and second in the
3461 directory in which the "foo.exe" file is located.
3463 Previous releases of OAFW would improperly return
3464 STATUS_NOT_A_DIRECTORY instead of the expected
3465 STATUS_OBJECT_PATH_NOT_FOUND. This would cause the Windows SMB client
3466 to terminate the search for the DLL causing the execution of the
3467 application to fail.
3469 * Rework the reference counting of smb_vc_t objects. The references
3470 stored in the waiting locks were not counted. This could result in
3471 an assertion if the reference count drops to zero.
3473 * "fs wscell" returns "Freelance.Local.Root" for the workstation cell
3474 when running in freelance mode.
3476 * Added support for persistent caching of file contents, stat entries,
3477 volume data, cell data, and name lookup data. When the registry
3478 value, "NonPersistentCaching", is not defined or set to 0 the file
3479 specified by "CachePath" becomes a persistent cache file. The
3480 size of the cache file is computed at run time. It has a maximum
3481 size of approximately 1.2GB. The contents of the file will be
3482 validated according to the rules specified by the "ValidateCache"
3485 * Added support for UUIDs. UUIDs are kept for as long as the
3486 cache file is intact.
3488 * Added cmdebug.exe and support cache manager debugging callback
3489 interfaces in afsd_service.exe.
3491 Usage: cmdebug -servers <server machine> [-port <IP port>] [-long]
3492 [-addrs] [-cache] [-help]
3493 Where: -long print all info
3494 -addrs print only host interfaces
3495 -cache print only cache configuration
3497 * Symlinks to \\AFS\[all\]... will now be treated the same as
3498 symlinks to /afs/.... However, please use /afs/... as otherwise
3499 the symlinks won't work on Unix.
3501 * Correct a problem with local CellID allocation for cells whose
3502 server list is obtained via DNS instead of a CellServDB file. If
3503 the DNS information expires the CellID assigned to the entry will
3504 be changed. This causes all of the cm_scache_t objects which refer
3505 to the old cellID number to become useless. Attempts to access
3506 files or directories with cm_scache_t objects using the old cellID
3507 will fail since the server list cannot be obtained.
3509 * Correct deadlock condition in cm_EndCallbackGrantingCall which can
3510 be triggered if the volume referenced by the cm_scache_t object is
3513 * The AFS Service needs to respond to SERVICE_ACCEPT_SHUTDOWN messages
3514 in addition to SERVICE_ACCEPT_STOP.
3516 * Move RPC shutdown until after the SMB and RX shutdown procedures
3517 complete. Block until RPC shutdown is complete.
3519 * Modify afslogon.dll (integrated logon) to wait for service if its
3520 state is START_PENDING. If the timeout period occurs, reset to
3521 the retry interval and not the sleep interval.
3523 * When renewing the server list for a cell obtained via DNS AFSDB
3524 records, the cm_cell_t entry must be removed from the list of all
3525 cells. Otherwise, the list of cells will be corrupted.
3527 * In the dcache and scache modules, use the cm_scache_t dataVersion
3528 instead of the cm_fakeDirVersion.
3530 * Synchronize fs.c with the unix version.
3532 * The variable used to determine whether a file or virtual memory
3533 mapped cache is used was not properly initialized to a default
3534 value. If the registry setting "NonPersistentCaching" was not
3535 set, the choice would be random. Properly initialized to be
3538 * The memory mapped view was never unmapped before closing the file
3539 at service shutdown. This is now properly cleaned up.
3541 * Default location of Cache file is now %TEMP%\AFSCache
3545 * A bug affecting new installations of 1.3.75/76 would result in
3546 the creating of incorrect mountpoints in the freelance root.afs
3547 volume for the default cell.
3549 If "fs \\afs\all\<cellname>" lists a volume name of "root" instead
3550 of "root.cell", you have been affected by the bug. To correct the
3551 problem, execute the following commands:
3553 fs rmmount \\afs\all\<cellname>
3554 fs rmmount \\afs\all\.<cellname>
3556 fs mkmount \\afs\all\<cellname> root.cell <cellname>
3557 fs mkmount \\afs\all\.<cellname> root.cell <cellname> -rw
3561 * A bug has been fixed in the auto-generation of Freelance root.afs
3562 symlinks which produced random entries in the \\AFS\all directory.
3564 * Support has been added for multi-homed servers
3567 * Added a new registry value, "StoreAnsiFilenames", which can be used
3568 to force the use of ANSI character sets instead of OEM Code Pages.
3569 This feature is useful when users require the ability to create
3570 filenames with 8-bit characters and need to access the files from
3571 both Latin-1 based Unix systems as well as from Windows.
3573 Activation of this feature will prevent access to files stored with
3574 8-bit OEM characters.
3576 * Shutdown all SMB threads in a synchronized manner when stopping the
3579 * There is currently a maximum cache size of 1.3GB. The limit is imposed
3580 by the largest contiguous block of unused memory within the 2GB process
3581 space which can be assigned to the memory mapped file. Unfortunately,
3582 when the executable digital signature verification code is activated
3583 Windows sees fit to further segment the process memory which in turn
3584 reduces the size of the maximum cache file to less then 800MB. If
3585 larger cache sizes are desired, a new registry value should be set:
3587 HKLM\SOFTWARE\OpenAFS\Client (DWORD) "VerifyServiceSignature" = 0x0
3589 Setting this value will disable the runtime verification of digital
3590 signatures on afsd_service.exe and the afs dlls which it loads. It
3591 will not disable the the version number check on those same files.
3592 The signature verification is not a security messure and is only meant
3593 to enhance the ability to afsd_service.exe to detect potential
3594 destablizing mixtures of DLLs from incompatible distributions.
3596 Added code to auto-disable the signature verification check if
3597 the desired cache size is greater then 700MB.
3599 * Windows' WinTrustVerify(WIN_SPUB_ACTION_PUBLISHED_SOFTWARE) is
3600 used to verify the validity of the afsd_service.exe binary
3601 as well as each of the AFS DLLs loaded by the service. Not only
3602 must the digital signature be valid but the signatures of the
3603 DLL must be signed by the same entity as the service.
3605 * Implement new functions: cm_freelanceMountPointExists and
3606 cm_freelanceSymlinkExists. Use them along with other validity
3607 checks in cm_freelanceAddMount and cm_freelanceAddSymlink to
3608 ensure that name collisions do not occur and that empty strings
3609 are not valid file names.
3611 A symlink may not have a name which would resolve to a valid
3612 cell name. Doing so would prevent access to the cell.
3614 * Add missing cm_HoldSCacheNoLock call to Freelance mount point
3615 re-initialization code. The reference counts of the fake root.afs
3616 volume scache object(s) would become invalid when the mount point
3617 or symlink lists were altered.
3619 * Add registry entries to provide mappings from the afsdsbmt.ini
3620 to the new locations for applications which count on the use
3621 of the old Profile file APIs. These apps are likely to fail
3622 if the user does not have administrator privileges and the
3623 registry is locked down.
3625 * The afs_config.exe submounts dialog had two errors.
3626 First, attempts to remove entries failed because the registry
3627 key was being opened without KEY_WRITE privileges.
3628 Second, when editing a submount entry, changing the name
3629 would add a new key and leave the original one in place.
3630 Now the original submount will be removed if its name is
3633 * In recent months there have been several incidents in which
3634 users have experienced problems starting or accessing
3635 afsd_service.exe and after significant effort has been spent
3636 it has turned out that they have two versions of AFS on the
3637 machine or an inconsistent set of DLLs.
3639 Code has now been added to afsd_service.exe which will walk
3640 the list of modules loaded by afsd_service.exe and validate
3641 that the version of the AFS DLLs matches the version of the
3642 afsd_service.exe executable. If they do not match the service
3645 * When Freelance mode is enabled and there is no registry
3646 key HKLM\SOFTWARE\OpenAFS\Client\Freelance, afsd_service.exe
3647 will attempt to import the afs_freelance.ini file contents.
3648 If the file does not exist, it was creating a dummy file
3649 with a r/o and r/w entry for the default cell and then
3650 importing those values.
3652 This process has been changed. The temporary file is no
3653 longer created. Also, both the OpenAFS Client install
3654 directory as well as %WINDIR% are checked for previous
3655 afs_freelance.ini files.
3657 * Added support for VL_GetEntryByNameN(). Still need to add
3658 support for VL_GetEntryByNameU() for multi-homed support.
3660 * Fix a deadlock situation in afscreds.exe when canceling an
3661 auto-generated Obtain Tokens dialog
3664 * Fix the locking of objects during Directory Searches in the
3665 SMB/CIFS server. The failure to properly lock the reference
3666 counts was resulting in the premature freeing of smb_dirSearch_t
3667 objects while they were still in use by the SMB/CIFS client.
3668 This does not solve the "Invalid Handle" problem.
3670 * Fix Find Cell By Name pioctl call to return a valid cell
3671 name for the Freelance fake root.afs volume.
3672 "Freelance.Local.Root".
3674 * Fix the Explorer Shell Extension Symlinks->Add operation.
3675 The dialog template was missing and the link destination
3676 string was too short.
3678 * Add support for symlinks to Freelance root.afs volume
3679 Stored at HKLM\SOFTWARE\OpenAFS\Client\Freelance\Symlinks
3680 <number> = "<linkname>:<relative-path>."
3681 Use symlink.exe to create, list, or remove
3683 * Remove the fallback to the use of KFW's KRB4 library when obtaining
3684 tokens. We never obtain KRB4 tickets.
3686 * Fix AFS Client Configuration Control Panel to support new SysName
3689 * Fix a bug in afsd_service.exe which could result in the SysName
3690 not being read from the registry.
3693 * Add code to block the issuance of AFS tokens by aklog.exe or
3694 afscreds.exe when the Kerberos 5 principal name contains a dot.
3696 * Modify the IsAdmin() function to always treat the local SYSTEM
3697 account as an AFS client administrator. Affects fs.exe and
3700 * Modify the internal handling of Quota Exceeded errors
3702 * Upgrade all reference count fields in the Windows cache manager
3703 and the osi library to use unsigned long instead of signed short.
3704 A similar fix has been applied to the afs rpc (rx) library.
3706 * fix the Windows cache manager to prevent it from replacing the
3707 rx_connection object associated with the cm_conn_t object on each
3708 and every operation if "fs crypt" was set. This explains the
3709 dramatic performance difference when crypt is used vs clear.
3710 The problem: 'cryptall', a boolean flag indicating whether or not
3711 "fs crypt" is set, was being compared to the rx_connection
3712 cryptlevel which is either rxkad_clear:0 or rxkad_crypt:2.
3713 1 != 2 and therefore the rx_connection was always destroyed
3714 and replaced on each and every operation.
3716 Lock the cm_conn_t object around every call to RXAFS_xxxx functions.
3717 It is not safe for the cm_conn_t object to not be locked because
3718 rx_DestroyConnection might be called from another thread if:
3719 - the user's tokens have changed or expired
3720 - the crypt mode has changed
3722 This fix appears to have also taken care of the problems associated
3723 with Overlapped Writes resulting in Delayed Write errors.
3725 * fix NSIS installer's AdminGroup.exe to properly create and
3726 remove groups when given -create or -remove. The string comparison
3729 * fs sysname now accepts a list of sysname values
3731 * added a new registry value HKLM\SOFTWARE\OpenAFS\Client "IoctlDebug"
3732 DWORD which when set to a non-zero value will cause error message
3733 text to be output to stderr from the pioctl() routine. Useful in
3734 debugging failures of fs.exe, tokens.exe, etc.
3736 * added a test to the power management code to only perform a
3737 flush operation if there is at least one network adapter which
3738 is not a loopback adapter.
3740 * Fix bug in loading of registry value HKLM\SOFTWARE\OpenAFS\Client
3741 "EnableKFW". This value will not be read if the key
3742 HKCU\SOFTWARE\OpenAFS\Client exists; even if the "EnableKFW"
3743 value under that key does not.
3745 * provide mechanisms to force the use of krb524d for Kerberos 5
3746 ticket to AFS token conversion. For afslogon.dll and afscreds.exe
3747 there is a new registry value "Use524" and for aklog.exe a new
3748 command line parameter "-m".
3750 * Fix the pattern matching algorithm to properly match patterns
3753 * smb_ReceiveCoreRename() was factored to produce smb_Rename()
3754 which is used by both the original function and the new
3755 smb_ReceiveNTRename(). smb_ReceiveNTRename() supports the
3756 creation of HardLinks in addition to Renaming. smb_Link()
3757 is a new function which creates HardLinks via cm_Link().
3758 cm_Link() is a new vnodeops function which creates links
3761 smb_ReceiveNTRename() does not support the File Copy and
3762 Move Cluster Information operations described in its interface.
3763 ReceiveNTRename is under documented in CIFS-TR-1p00_FINAL.pdf.
3765 * When opening files via symlinks, we should follow the symlinks
3766 until we reach the actual file stat cache entry. The stat cache
3767 entry of the file should then be stored in the FID instead of
3768 stat scache entry of the symlink.
3770 * return bad operation errors for all unimplemented functions
3771 even if we do not know the functions exist.
3773 * Log bad packets and unknown operation packets to the trace log
3775 * Map CM_ERROR_BADOP to STATUS_NOT_SUPPORTED instead of
3778 * Update list of known CIFS operations to include all those listed
3779 in CIFS-TR-1p00_FINAL.pdf.
3781 * Modify the handling of HKLM\SOFTWARE\OpenAFS\Client\Submounts
3782 to support the REG_EXPAND_SZ type.
3785 * A new Windows authorization group "AFS Client Admins" is now
3786 created and populated with the members of the "Administrators"
3787 group. The group is used to determine which accounts on the
3788 machine may be used to modify the AFS Client Configuration via
3789 the UI and command line tools. afs_config.exe, fs.exe,
3791 * Modify the WinLogon Logoff Event Handler to query NT4 domain
3792 controllers for the remote profile path if Active Directory
3793 services are not available.
3795 * Fix aklog.exe to not add the AFS ID to the username
3797 * PTS registration of new users to foreign cells has been added to
3800 * The cm_Daemon thread is used to perform checks for
3801 down servers, up servers, volumes, callback expirations,
3802 lock maintenance and token expiration. Due to a gaff in
3803 larger integer division the thread never performed any
3804 work. Instead the current time computation would always
3805 be less then the trigger times. This had an adverse affect
3806 on the client's ability to maintain communication with servers,
3807 keep volumes up to date, and flush user tokens and acls
3808 when they have expired. This was broken when the 1.3 branch
3809 was modified to support VC7 which no longer included
3812 * An initialization problem with the Freelance code was
3813 detected while fixing the callbackRequest. The cm_rootSCachep
3814 object is obtained during afsd_InitDaemons() but the callback
3815 information is incomplete. The callback information will not
3816 be obtained until cm_MergeStatus is called from within
3817 cm_GetCallback. Unfortunately, cm_SyncOp did not properly
3818 test for the conditions under which the callback information
3821 * Reports have been filed indicating that callbacks were
3822 being lost. An examination of the code indicated that the
3823 cm_server_t objects were not being properly reference
3824 counted by the cm_scache_t and cm_callbackRequest_t objects.
3825 In particular, the cm_server_t objects may have been freed
3826 from beneath the cm_conn_t objects.
3828 All of the reference counting is now done via the functions:
3833 this improves the ability to track the referrals.
3835 Each cm_BeginCallbackGranting Call now allocates a reference
3836 to the cm_server_t. The cm_EndCallbackGrantingCall either
3837 frees the reference or transfers it to the cm_scache_t
3838 cbServerp field. These are then appropriately tracked
3839 through the cm_Analyze call.
3841 * Ensure that the dnlc hash table is the same size as the
3842 dir name hash table (as per original author's note).
3843 Increase the dnlc CM_AFSNCNAMESIZE to a multiple of 8
3844 for compatibility with 64-bit systems.
3846 * fix smb_ApplyV3DirListPatches to properly apply the hidden
3847 attribute to dotfiles when the infoLevel < 0x101 and
3848 cm_SyncOp has failed.
3850 * Fix the Freelance registry initialization code. There
3851 was a possibility that some systems could end up with
3852 garbage in the registry during a clean install.
3855 * file and directory names beginning with "." will now be given the
3856 hidden attribute when the volume access is anonymous. this matches
3857 the behavior when the volume access is via an authenticated user.
3859 * Added a change monitor to the HKLM\SOFTWARE\OpenAFS\Client\Freelance
3860 key. When a change occurs mark the root.afs data as invalid and
3861 for it to be reloaded on the next access. This allows administrators
3862 to modify the mount point list without restarting the service.
3864 The freelance client used to provide a fake modification time for
3865 the root.afs volume data and its mount points of 7/09/2001 14:24 EDT.
3866 Added code to extract the last modification time of the Freelance
3867 registry key and use that instead. The time now represents the
3868 most recent mount point change.
3870 * PTS registration of new users to foreign cells has been added to
3873 * Additional Cache Control and Credential Manager options have been
3874 added to the WiX installer. See deployment guide for details.
3876 * The CachePath setting is now optionally a REG_EXPAND_SZ type
3878 * The WiX installer has been upgraded. Version 2.0.1927.1 is now
3881 * The loopback installation code may have had a problem updating the
3882 %ETC%\HOSTS file which could have resulted in a premature failure.
3883 Work around code has been added for the case where the file cannot
3886 * The default max chunksize was increased from 15 (32K) to 17 (128K)
3887 because Windows sends 64K blocks when using overlapped writes.
3889 * The default number of server threads was increased from 4 to 25 to
3890 better handle overlapped writes.
3892 * The "AfscredsShortcutParams" registry value was not being properly
3893 loaded by afscreds.exe. Therefore, the default value was always being
3894 used instead of the value set by the installer.
3896 * Windows XP provides downgrade attack detection to prevent an attacker
3897 from being able to force the use of NTLM simply by disrupting
3898 communication with the KDC. This attack cannot exist between the
3899 Windows CIFS client and the AFS Client Service. Therefore, when a
3900 downgrade has been detected the afs pioctl library will force the
3901 establishment of a new CIFS connection using NTLM.
3903 * A locking error was discovered surrounding all references to volume
3904 server lists within the cm_cell.c source file.
3906 * The logged into Windows username was incorrect on Terminal Server
3909 * A new registry value "NonPersistentCaching" was added to the service
3910 parameters key. When set to a non-zero value, the afs cache is stored
3911 in the Windows paging file. There are two limitations to choosing
3913 1. when persistent caching is implemented it won't work with
3914 this flag set since there will be nothing to persist.
3915 2. with this flag set the initial paging allocation cannot be
3916 changed while the service is running
3918 * An initialization bug was discovered in aklog.exe which affected users
3919 who have a domain name for their afs servers which could not be mapped
3923 * afs_config.exe now validates cell names against DNS in addition
3924 to the CellServDB file.
3926 * In order to allow the freelance client to connect to a volume with ID
3927 equal to 1 on the default cell we changed the fake root.afs volume ID
3928 once again. This time we choose 0xFFFFFFFF. In addition, we change
3929 the cell ID of the fake root.afs volume from 1 to 0xFFFFFFFF as well.
3930 It will now be impossible for a volume ID to match that of another
3931 cell unless the client is connected to 0xFFFFFFFD cells. That should
3932 be enough room for growth.
3934 * Fix "fs mkmount" command to work with UNC paths and when
3935 started from non-AFS drives. It is now possible to create a mount
3936 point in the freelance fake root.afs volume with the command
3938 fs mkmount \\AFS\all\<directory-name> <volume-name> <cellname>
3942 fs mkmount \\AFS\all\openafs.org root.cell openafs.org
3943 fs mkmount \\AFS\all\.openafs.org root.cell openafs.org -rw
3945 * The algorithm used to re-attempt access to the servers associated with
3946 a volume has been altered to properly address the case in which all
3947 servers have been marked down. The previous algorithm did not reset
3948 the server's down flags so the servers were never actually retried.
3949 This caused a problem with active volumes if the network connectivity
3950 was lost as could be the case with a network cable removal, wireless
3951 drop, or laptop hibernation. With the fix volume access is restored
3952 almost instantenously when network connectivity becomes available.
3954 * Support for SMB/CIFS browsing has been added to the AFS Client Service
3955 SMB server. It is now possible to use "NET VIEW \\AFS" to obtain a
3956 listing of AFS submounts and freelance mount points. Support for
3957 NETSHAREENUM, NETSHAREGETINFO, NETSERVERENUM2, NETSERVERGETINFO
3958 significantly enhances the behavior of AFS volumes within the Explorer
3959 Shell. For instance, "AFS" now shows up as server in the Explorer
3960 with each submount or freelance mount point visible as a share.
3961 The right click menu in each folder now works with full functionality
3962 on a consistent basis.
3964 * The network provider can be configured to have different behavior
3965 depending on the domain that the user logs into. These settings are
3966 only relevant when using integrated login. A domain refers to an
3967 Active Directory (AD) domain, a trusted Kerberos (non-AD) realm or the
3968 local machine (i.e. local account logins). The domain name that is
3969 used for selecting the domain would be the domain that is passed into
3970 the NPLogonNotify function of the network provider. (see registry.txt
3973 * Added a new registry value [HKCU\SOFTWARE\OpenAFS\Client]
3974 "Authentication Cell" which may be used to specify a default
3975 authentication cell for afscreds.exe which is different from
3976 the default cell for the AFS Client Service daemon.
3978 * Added a Logoff WinLogon Event Notification function to afslogon.dll.
3979 afslogon.dll moved to %WINDIR%\System32\.
3980 New registry entries added to register the dll for Winlogon events.
3982 The logoff event will now force a call to ktc_ForgetAllTokens()
3983 using the context of the user being logged off as long as the
3984 user's profile is not loaded from within AFS. If the profile
3985 was loaded from AFS we can't release the tokens since the Logoff
3986 event is triggered prior to the profile being written back to
3987 the its source location. This is now performed in an XP SP2
3990 * Windows XP SP2 Internet Connection Firewall interoperability
3993 * The %WINDIR%\afsdsbmt.ini contains four sections:
3994 Submounts, Drive Mappings, Active Maps and CSC Policies.
3995 The Submounts and CSC policies are now stored in the registry under
3996 [HKLM\SOFTWARE\OpenAFS\Client\Submounts]
3997 [HKLM\SOFTWARE\OpenAFS\Client\CSCPolicy]
3998 The Drive Mappings and Active Maps are stored in the registry under
3999 [HKCU\SOFTWARE\OpenAFS\Client\Mappings]
4000 [HKCU\SOFTWARE\OpenAFS\Client\Active Maps]
4002 There is no automatic migration of this data as it would be impossible
4003 to consistently migrate data to user profiles which may not be active
4004 when the machine is updated.
4006 * The %WINDIR%\afs_freelance.ini contains lists of mountpoints for the
4007 fake root.afs volume. For the same reasons as for the cellservdb file,
4008 this information should not be in %WINDIR%. This information is now
4009 kept under the registry key
4010 [HKLM\SOFTWARE\OpenAFS\Client\Freelance]
4012 The data from the afs_freelance.ini file will be automatically
4013 migrated to the registry on first execution of afsd_service.exe
4015 * Keeping the CellServDB file in the location %WINDIR%\afsdcell.ini is
4016 troublesome for several reasons. One, it is confusing for those who
4017 expect the file to be named "CellServDB" instead of "afsdcell.ini".
4018 Two, this file is not a Windows Profile formatted file. Three,
4019 applications should not be reading or writing to %WINDIR%. It causes
4020 problems for Windows Terminal Server.
4022 The new location of CellServDB will be the OpenAFS Client install
4023 directory which is by default C:\Program Files\OpenAFS\Client and can
4024 be determined by querying the registry for
4025 [HKLM\SOFTWARE\TransarcCorporation\AFS Client\CurrentVersion]PathName
4027 The existing afsdcell.ini will be migrated by the NSIS installer.
4028 The Wix installer must still be updated to do the same.
4030 * Change NSIS installer to use DNS by default; to remove Integrated Logon
4031 High Security mode; and to add Terminal Services compatibility registry
4032 entries to allow the OpenAFS tools to find the afsdcell.ini and other
4033 configuration files in %WINDIR%.
4035 * Add support for authenticated SMB connections. This will remove
4036 the need for high security mode in most situations. Both NTLM
4037 and Extended Security (GSS SPNEGO) modes are supported. Effectively,
4038 only NTLM can be used even though Kerberos is now supported. The
4039 reason is that it is not possible to construct a service principal
4040 which is unique to each individual machine.
4042 SMB Extended Auth does not work on XP SP2 unless one of two registry
4043 modifications are made:
4045 (1) To disable the check for matching host names on loopback connections
4046 set this key. This does not require a reboot:
4048 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa]
4049 "DisableLoopbackCheck"=dword:00000001
4051 (2) To add the AFS SMB/CIFS service name to an approved list. This
4052 does require a reboot:
4054 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa\MSV1_0]
4055 "BackConnectionHostNames"=multi-sz "AFS" "MACHINE-AFS"
4057 afsd_service.exe will automatically add the current Netbios Name
4058 to the BackConnectionHostNames list and then temporarily disable
4059 the loopback check for one cycle of startup/shutdown of the service.
4060 We assume most folks do not start/stop without a reboot so this
4061 will be adequate in most cases.
4063 * Fix security hole in afslogon.dll which allowed passwords to be
4064 sent in clear text to the KDC in a misformed principal name.
4066 * Fix cm_GetCell() to properly handle expired dns entries
4069 * If Freelance mode is active and the afs_freelance.ini
4070 file does not exist, do not create an empty file.
4071 Instead create a file containing ro and rw mountpoints
4072 to the default cell using the standard conventions.
4074 * Modify the Freelance support to handle the ability
4075 to create rw mount points in the fake root.afs volume.
4077 * Changed the RPC mechanism used for token setting from
4078 named pipes to local. Use of named pipes can be restored
4079 by setting the environment variable AFS_RPC_PROTSEQ to
4082 Named pipes were required when a Windows 9x system was
4083 using a NT system in gateway mode which is incompatible
4084 with our use of local loopback adapters.
4086 * In afscreds.exe, if a username of the form user@REALM is
4087 specified and no password is specified, do not perform a
4088 kinit operation. Only perform the aklog functionality.
4090 * Add a new registry value which allows the number of processors
4091 on which afsd_service.exe executes to be restricted. Valid
4092 values are 1..numOfProcessors
4094 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
4098 * A second MSI based installer option is now available.
4100 * Fixed Kerberos 5 kinit functionality in afscreds.exe to properly
4101 request tickets for user/instance@REALM instead of just user@REALM
4103 * Modify the Power Management Notify routine to wait for the Hard Dead
4104 timeout period instead of a fixed 19 seconds. With the longer timeout
4105 periods Hibernation and Standby could never succeed when network
4106 connectivity is not available.
4108 * The following fs.exe commands are now restricted to Administrator:
4109 - checkservers with a non-zero timer value
4112 - sysname with a new sysname list
4121 setting the default sysname for a machine should be done via the
4122 registry and not via "fs sysname".
4124 * NSIS installer adds options to install Debugging Symbols
4125 and the Microsoft Loopback Adapter; the user is now also
4126 given the ability to select the afscreds.exe startup options.
4128 * Build system modified to generate symbols for FREE (aka RELEASE)
4129 builds as well as CHECKED (aka DEBUG) builds
4131 * Sites which have a volume ID of 0x20000001 assigned to their
4132 root.cell volumes have been experiencing problems with accessing
4133 the root.cell volume of their cell when Freelance mode has been
4134 active. This was because 0x20000001 was assigned to the fake
4135 root.afs volume created by freelance. The fake volume id is
4136 now set to 0x00000001 to prevent conflicts.
4138 * The timeout logic in the AFS Client Service has been wrong
4139 for sometime. It is based on two different assumptions.
4140 First, the SMB client timeout is a fix value as was the case
4141 with OS/2 Lan Manager. This assumption is incorrect. The
4142 SMB timeout in Windows is a dynamic value computed based upon
4143 a fixed minimum timeout to which is added time based upon the
4144 size of the request and the performance characteristics of
4145 the connection. Second, it is the responsibility of the
4146 SMB Server to enforce the timeout requirements of the client.
4147 This is untrue. The SMB Server cannot be expected to know
4148 the requirements of the client. More importantly, if the
4149 SMB server uses the SMB client timeout as a value to restrict
4150 its behavior as an RX client, the performance characteristics
4151 of the local SMB session would be used to prematurely terminate
4152 WAN connections with significantly different performance
4155 The timeout logic has therefore been modified in the following
4157 . the Lan Manager Workstation (SMB) Session Timeout is used only
4158 as a basis for configuring the Connection Dead Timeout
4159 and Hard Dead Timeout values. The Connection Dead Timeout
4160 must be at least 15 seconds longer than the SMB Timeout
4161 and the Hard Dead Timeout must be at least double the
4162 Connection Dead Timeout.
4163 . New registry entries have been added to allow the Connection
4164 Dead Timeout and Hard Dead Timeout values independent of the
4165 Lan Manager Workstation Session Timeout
4166 . The test to enforce the SMB Client Timeout has been removed.
4168 One of the side-effects of removing the enforcement of the SMB
4169 Client Timeout is that regardless of whether or not the SMB client
4170 is available to receive the response (and how would the SMB server
4171 know) the RX protocol response can be used to update the AFS
4172 Client Service state for ready access by future SMB client
4175 This should be the end of the "Server paused or restarting messages"
4177 * Add support for arbitrary UNC paths to the pioctl() support.
4178 This enables the fs commands as well as the AFS Shell Extension
4179 to work correctly when UNC paths are being used.
4181 * Fix afscreds.exe (by updating afskfw.lib) to search for cells via
4182 DNS if the cell configuration cannot be determined via CellServDB
4184 * Add debug info to test whether CM_BUF_WAITING or CM_SCACHE_WAITING
4185 are ever set more than once at a time
4187 * Fix the management of lists of cm_cell_t structures when using
4188 DNS to lookup cell information. The previous code would fail to
4189 reuse the same cellID for a cell if DNS was used more than once
4190 for a given cell name. When the ttl expired, a single cm_cell_t
4191 could be inserted into the cm_allCellsp list more than once
4192 producing a loop. In addition, the vlServerp list belonging to
4193 the cell was not freed resulting in improper refCounting of the
4196 * Add DNS support to cm_IoctlNewCell() which previous only examined
4199 * Add cm_FreeServer() function and call it from cm_FreeServerList()
4200 to properly garbage collect cm_server_t objects
4202 * Add numVCs variable to smb.c to track the number of smb_vc_t
4203 objects created and use it to initialize the vcID field which
4204 previously was set to 0 in all objects resulting in FindByID
4207 * Fixed DNS lookups to work consistently throughout the OpenAFS
4208 product instead of just from within the afsd_service.exe
4210 * Added a runtime check to ensure that AFS Client Service SMB
4211 Server is accepting connections before attempting to mount
4214 * Read IP addresses for volume servers out of the CellServDB
4215 file if gethostbyname() on the hostname fails.
4217 * Fix getcellconfig() to populate both the Hostnames as well
4218 as IP addresses when loading cell data via DNS
4220 * Increase the Connection Dead Time to 50 from 20 seconds
4221 Increase the Hard Dead Time to 120 from 40 seconds
4222 (matches the Unix values)
4224 * Fixed an assertion validating the number of allocated NCBs
4226 * Fixed the build environment to consistently build for
4227 Windows 2000 and above. (APPVER = 5.0)
4229 * Fixed rx_debug to properly validate the receipt in incoming
4230 data with select() and recvfrom(). Do not copy data out of
4231 the socket buffer unless success is indicated.
4234 * afsd_service.exe will now display a message box to the
4235 desktop when it terminates due to an IP Address Change.
4237 * installer no longer deletes AFS Server configuration data
4240 * installer generates a warning dialog if the RPC service
4241 is not properly configured
4243 * installer compressed with lzma instead of bzip2
4245 * afsd_service.exe shutdown crash solved once and for all
4247 * reference counting of smb_vc_t data structures improved
4249 * name space collision of smb_fid_t event objects corrected
4251 * the output of "fs memdump" is now written to
4252 %WINDIR%\TEMP\afsd_alloc.log
4254 * the file TaAfsApp_1033.dll is now properly installed allowing
4255 the User Manager to start
4257 * a new algorithm is used for computing filename pattern matches
4259 * afscreds.exe now accepts user names containing instance
4262 * Fix the Directory Name Lookup Cache to be case-sensitive.
4263 This is crucial in environments in which a Windows client
4264 is accessing a directory with more than one filename that
4265 differs only by case. If the directory contains "FOO"
4266 and "Foo". You want "DEL Foo" to delete the correct one.
4267 We still have a problem in that "DEL foo" will delete a
4268 random filename. This will be addressed in a future release.
4270 * Fix afscreds.exe -M option (renewMaps) to work when High
4271 Security mode is off. Also, remember to disable the ActiveMap
4272 flag in afsdsbmt.ini when a drive mapping is removed.
4274 * Updates to NSIS installer script. AFS Server configuration
4275 data will not be destroyed on un-install or re-install.
4276 Use a better compression algorithm.
4278 * afslogon.dll now uses KFW to obtain tokens when available
4280 * afslogon.dll when given an all uppercase username will
4281 attempt to authenticate with both the uppercase name
4282 and an all lowercase variation
4284 * DST modification removed. The fix appears to make things
4285 worse after a reboot of the machine.
4287 * fs.exe: added "cscpolicy" which is used to
4288 change client side caching policy for AFS shares
4290 Usage: fs cscpolicy [-share <AFS share>] [-manual] [-programs]
4291 [-documents] [-disable] [-help]
4293 * Several uninitialized variables have been initialized
4295 * It is now possible to obtain tokens using cross realm
4296 Kerberos within afscreds.exe:
4298 user: jaltman@ATHENA.MIT.EDU
4300 Will obtain a cross realm ticket for jaltman/DEMENTIA.ORG@ATHENA.MIT.EDU
4301 will will in turn be used to obtain afs@DEMENTIA.ORG.
4302 The resulting token will be stored with the display name
4303 jaltman@ATHENA.MIT.EDU@dementia.org
4305 * aklog.exe has been added to the client
4307 Usage: aklog [-d] [[-cell | -c] cell [-k krb_realm]]
4308 [[-p | -path] pathname]
4312 -d gives debugging information.
4313 krb_realm is the kerberos realm of a cell.
4314 pathname is the name of a directory to which you wish to authenticate.
4315 -noprdb means don't try to determine AFS ID.
4316 -5 or -4 selects whether to use Kerberos V or Kerberos IV.
4317 (default is Kerberos V)
4318 No commandline arguments means authenticate to the local cell.
4321 * All of the resource files have been restructured to adhere to
4322 a set of rules IBM implemented for loading string resources.
4323 These rules had either been forgotten or were not discovered
4324 by folks working on the OpenAFS sources. The end result was
4325 memory corruption. This is primary item which was preventing
4326 the AFS Server from working.
4328 * Increased the size of the maximum ticket size stored in a token
4329 from 344 bytes to 12,000. Increased the buffers used to convey
4330 messages between the pioctl() caller and the SMB Server from
4331 1000 bytes to 12,512. The code appeared to have been writing
4332 above the top of the stack by quite a few number of bytes.
4333 (The increased ticket size is necessary for the next item.)
4335 * When obtaining AFS Tokens via KFW, krb524 is no longer required.
4336 Instead the raw Kerberos 5 ticket is used in its entirety. This
4337 is extremely important as it allows us to use pure Kerberos 5 KDCs
4338 as the source of the AFS authentication. The use of up to 12,000 byte
4339 tickets will allow tickets produced by all versions of Microsoft
4340 Active Directory to be used.
4341 - create a user account.
4342 - designate it DES only
4344 - specify its UPN to be "afs@realm"
4345 - assign a SPN of "afs/cellname" to the UPN with setspn.exe
4347 * Do not enforce the funky 8dot3 pattern matching rule that the first "."
4348 is special when using long file names. (you must use "*.*" and not "*")
4349 Instead only enforce it when performing 8dot3 searches.
4351 * Fixed the DST problem with creation times being set one hour ahead
4353 * Fixed the problem when using \\afs\cell-alias. For example,
4354 \\afs\uncc instead of \\afs\uncc.edu. Do not a new cell struct
4355 for the alias name; instead simply expand the name. One of the
4356 symptoms of this problem was a loss of acquired tokens.
4358 * Fixed the AFS Shell Extension. The Symbolic Link menu was empty
4359 of strings. (Only English strings provided.)
4361 * Fixed the installer to properly replace in use files.
4363 * Fixed the build system to cleanup generated component version files
4365 * The release build compiled with MSVC 6.0 compiler to avoid the
4366 afsd_service.exe shutdown crash. This does not solve the problem
4367 but simply avoids it for the time being.
4371 * fix afslogon.dll to not corrupt memory when High Security mode
4374 * fix afsd_service.exe to not attempt to restore the stack when
4375 an exception occurs. (not safe in multi-threaded programs)
4377 * fix uninstaller to properly remove the CRT and MFC DLLs
4379 * remove a Message Box from afscreds.exe when getcellconfig()
4380 fails on a kerberos realm which is not a cell
4382 The following is a list of changes to the OpenAFS for Window client
4385 * "fs setserverprefs" will leave afsd service deadlocked
4387 * "vos listaddrs" will core dump
4389 * installer sets the appropriate keys to support Integrated Logon
4391 * installer disables the "Find Lana by Name" functionality as it
4392 was causing headaches for many users
4394 * fix the intermittent crash of the power management thread when
4395 shutting down the AFS Client Service
4397 * optimizes the obtain drive mount list functionality which is
4398 executed every time the mount tab in afscreds.exe and afs_config.exe
4399 are refreshed. (this happens a lot)
4401 * fix the service shutdown logic. add the STOP_PENDING state
4402 and do not accept additional service events after we declare
4405 The following is a list of changes to the OpenAFS for Window client
4408 * flexelint was run against the source tree and hundreds (perhaps
4409 thousands) of corrections were applied to ensure prototypes
4410 were in use; types were used consistently; variables were
4411 initialized; unused variables were removed; etc.
4413 * A wide variety of instrumentation was added including the
4414 ability to produce a stack trace from within afsd_service.exe
4417 * Dynamic configuration of the RDRtimeout value based upon the
4418 LanMan Workstation Session Timeout
4420 * The mount root no longer needs to be called "/afs". This
4421 is now set by a registry value "MountRoot" within the key
4422 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
4424 * The cell list is now only read out of afsdcell.ini when the
4425 file changes instead of each time a cell is resolved.
4427 * Thread synchronization was added to cm_server.c and ktc_nt.c
4429 * All calls to GlobalAlloc()/GlobalFree() were replaced with
4430 calloc()/free(). The Global functions were needed on Windows 3.x
4431 but have caused a variety of problems on the Win32 platforms.
4432 Avoiding them is highly recommended by several Microsoft
4433 Knowledgebase articles
4435 * Support for Symbolic Links added to the AFS Shell Extension
4437 * Added a registry value "OverlayEnabled" to determine if
4438 Shell Extension Overlays should be enabled.
4439 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
4441 * New Build system to support VC6, VC.NET, VC.NET2003 compilers and
4442 separate trees for checked and free builds. Build system supports
4443 a custom directory src\WINNT\extra which can be used as a grafting
4444 location of organization specific additions to the build tree.
4446 * New installer built using NSIS 2.0.
4448 * Named all kernel objects in order to allow them to be monitored
4449 with tools such as SysInternals' ProcExp.exe.
4451 * Introduced new EventLog framework for AFSD
4453 * Introduced Power Management interface to AFSD for Standby and
4454 Hibernate modes to allow cache to be flushed prior to network
4457 * Utilize Win32 DNSQuery API instead of internal routines. This
4458 allows DNS SRV queries to be sent to all current domain name
4459 servers. Not just one specified in an INI file. DNS is now
4462 * "NetbiosName" registry value may be used to specify a fixed
4463 Netbios Name such as "AFS" to be used instead of "HOSTNAME-AFS"
4464 when the loopback adapter is in use. If you need to use the
4465 old notation with a loopback adapter installed specify a registry
4468 "NetbiosName" REG_EXPAND_SZ = "%COMPUTERNAME%-AFS"
4470 * Refactor all modules which depend on LAN Adapter and NetbiosName
4471 determination in a new library: lanahelper.lib. This allows for
4472 consistent behavior throughout the product.
4474 * Move the afsd.log and afsd_init.log files to the directory specified
4475 by the "TEMP" environment variable. This is usually %WINDIR%\TEMP
4476 for services. Added the Date to the log entries.
4478 * New registry value "RxMaxMTU" used to limit the size of the RX
4479 packets sent by the AFS Client Service to the Server. In order
4480 to enable OpenAFS to work across the Cisco IPSec VPN the packet
4481 size must be restricted to 1264 or smaller. The latest NSIS
4482 installer sets a value of 1260 by default.
4484 * New registry value "RxNoJumbo" to disable the use of Jumbo Rx
4485 packets. This is not needed in order to work across the Cisco
4486 VPN but might be needed for other network environments. This
4487 value is not set by the NSIS installer.
4489 * New registry value "HideDotFiles" is used to apply the Hidden
4490 attribute to files whose names begin with a '.'. This value
4491 is set by the NSIS installer.
4493 * New registry value "MaxMpxRequests" allows the maximum number
4494 of multiplexed sessions to be configured at run time. This
4495 value is not set by the NSIS installer. The default value is
4498 * New registry value "MaxVCPerServer" allows the maxmimum number
4499 of VCs per server to be configured at run time. This value is
4500 not set by the NSIS installer. The default value is 100.
4502 * New registry value "AllSubmount" allows the "all" submount to
4503 be disabled by setting its value to 0x00.
4505 * Allow cells names to be valid mount points
4506 \\<netbiosName>\<cellname>
4508 * Store the active state of drive mappings in order for afscreds.exe
4509 to restore them upon startup
4511 * Add exception handling to generate a Stack Trace to the afsd_init.log
4512 file if one happens to occur.
4514 * Add lots of logging to help detect the cause of invalid SMB packets
4516 * Enable Kerberos for Windows to be used to obtain AFS Tokens via
4517 conversion of Kerberos 5 "afs" service tickets. Supports auto-
4518 renewal of expiring tokens as long as afscreds.exe is running.
4520 * New afscreds.exe command line options:
4522 -M = renew drive maps
4523 -N = ip address change detection
4526 * New registry value "EnableKFW" in {HKCU,HKLM}SOFTWARE\OpenAFS\Client
4527 determines whether or not MIT Kerberos for Windows should be used
4528 to obtain tokens via Kerberos 5 tickets.
4530 * New registry value "AfscredsShortcutParams" in
4531 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
4532 determines the command line parameters to be specified when "fixing"
4533 the AFS Shortcut in the user's startup folder.
4535 * The "ShowTrayIcon" registry value has been moved from
4536 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
4537 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
4539 * The <cell name> registry values used to store the token expiration
4540 reminders have been moved from
4541 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
4542 {HKCU,HKLM}SOFTWARE\OpenAFS\Client\Reminders
4544 * Obtain the Logon User Name from the Explorer key when available
4546 * new text document doc\txt\winnotes\registry.txt lists all registry
4547 values used by OpenAFS (excluding the AFS Server)
4549 * BUG: rx_securityClass objects were not properly reference
4550 counted and were never freed.
4552 * BUG: reduce the number of conditions under which CM_ERROR_TIMEOUT
4553 would be generated. The existence of a server does not imply
4554 that it is not down. If all of the servers for a cell are down
4555 return CM_ERROR_NOSUCHVOLUME instead. This prevents the Explorer
4558 * BUG: the directory name lookup cache failed to free the entries
4559 in the cache when the name cache entries cycled. The entries
4560 in the cache would become dereferenced without being freed.
4562 * BUG: fs setserverprefs could be executed without Administrator
4565 * BUG: the number of allocated NCB objects (100) exceeded the number
4566 which could actually be waited upon by the kernel (64). Any objects
4567 which were utilized above the limit could never have event completions
4570 * BUG: smb_username_t objects were not being reference counted and
4571 were not properly freed.
4573 * BUG: smb_tid_t objects could under unusual circumstances be freed
4574 before they were no longer referenced.
4576 * BUG: smb_fid_t object pointer were frequently used even when
4577 their value could be NULL. They were not properly released and
4578 therefore they were never freed.
4580 * BUG: smb_packet_t data structures were not completely initialized
4583 * BUG: when Rx produces a CM_ERROR_NOIPC error do not return "Access
4584 Denied" because that causes the Explorer Shell to try again until
4585 access is obtained. Instead return "Remote Resources" which allows
4586 the shell to move on and treat the error as transient.
4588 * BUG: when initializing the NCBreturns structure, separate Event objects
4589 were created for each NCB although a single Event object was supposed
4590 to be shared by all.
4592 * BUG: smb_dirSearch_t objects were not being properly referenced counted
4595 * BUG: smb_tran2Packet_t objects were not being properly referenced
4598 * BUG: directory path creation did not handle the case of multiple
4599 directories requiring creation in one attempt
4601 * BUG: SMB requests which required an Extended Response were ignored.
4602 This prevented some files from being written to AFS volumes.
4604 * BUG: character strings were being freed even after they were
4605 inserted into in use data structures
4607 * BUG: inconsistent usernames were used when High Security mode was
4608 enabled. (there is still much to do in this area)
4610 * BUG: pioctl() calls which require out of band RPC operations were
4611 susceptible to race conditions when performed by multiple processes
4613 * BUG: memory allocation and deallocation crossed instances of the
4614 C Runtime Library producing memory leakage and corruption in
4615 afscreds and the client configurator.