1 Since 1.5.9902 (1.6pre2)
2 * rx calls issued on a busy call channel are now specially
3 handled and logged. All calls are automatically retried.
5 * The vos support for multi-homed servers was incorrectly
6 implemented. The changes have been reverted.
8 * Avoid wasting cm_buf_t objects and associated data buffers
9 for obtaining mountpoint and symlink target strings. Instead
10 read them directly into the cm_scache_t mountPointTarget
13 * Optimize the evaluation of mountpoint and symlink targets
14 by skipping the FetchStatus and a full round-trip if
15 the cache manager does not have valid status information
18 * Fix the abstraction of cm_FreelanceAddMount() by passing
19 in volume names that do not include the trailing dot.
22 [there was no 1.6pre1 for Windows]
24 * vos commands now manipulate servers by UUID and can
25 recognize multi-homed servers.
27 * afs_config will not longer set the Tray Icon State
28 in the registry if the checkbox is not present in
29 the dialog. [RT 128591]
31 * Heimdal's roken utility library has been added
32 as \Program Files\Common\afsroken.dll
34 * When probing servers to determine if they are up or down
35 no longer issue an RXAFS_GetTime RPC to servers that indicated
36 that they do not understand the RXAFS_GetCapabilities RPC.
37 Since they responded it is known that they are up.
39 * AFS Explorer Shell Extension now works from folder
40 backgrounds. Overlays for mount points and symlinks
41 are present in the dll, but are not registered at present
44 * Do not use RankServerInterval registry value as the value for
45 PerformanceTuningInterval.
47 * CellServDB updated to 13 Dec 2010 release from grand.central.org
49 * Add "fs chmod" command and display current mode as part
50 of "fs examine" output.
52 * When the data version of a mountpoint or symlink changes,
53 the target string in the cm_scache_t object must be cleared.
54 Otherwise, the new target will not be queried.
56 * "fs checkservers" now includes vldb servers in the output
57 and only lists multi-homed servers once. A multi-homed
58 server that has at least one up interface is no longer
59 considered to be down.
61 * When asynchronously storing dirty data buffers to the
62 file server ensure that (a) the cm_scache_t object and
63 the cm_buf_t object are for the same File ID so that
64 locking and signalling work properly; and (b) if the
65 FID no longer exists on the file server, do not panic,
66 just discard the buffer.
68 * When processing VNOVOL, VMOVED and VOFFLINE errors perform
69 server comparisons by UUID or address and not simply by
70 cm_server_t pointer. Otherwise, server failover may not
73 * Do not preserve status information for cm_scache_t objects
74 when the issuing server is multi-homed.
76 * Giving up all callbacks when shutting down or suspending
77 the machine is now significantly faster due to the use
78 of an rx_multi implementation. (This functionality is
79 still off by default and must be activated by a registry
82 * Race conditions were possible when updating the state
83 of the cm_volume_t flags and when moving the volumes
84 within the least recently used list.
86 * Ensure that the lanahelper library does not perform a
87 NCBRESET of each lan adapter when enumerating the
88 current network bindings. Correcting this permits OpenAFS
89 to work on Windows 7 when the network adapter settings
92 * Fix creation of mount points and symlinks as \\AFS\xxxx
96 * Reimplement FetchData and StoreData operations to make
97 use of rx_Writev and rx_Readv which is faster than than
98 multiple calls to rx_Write and rx_Read. Throughput
99 improvements of up to 20% are possible.
101 * Many improvements to the rx rpc stack:
105 - Fast recovery is not triggered by packet resend timeouts
107 - Socket errors are now detected and reported
109 - avoid repetitive memory allocations
111 - avoid calling gettimeofday() for every packet
113 - reduced dependency on the call->lock
115 - When a call receive is completed, ack all of the packets
116 immediately. Do not wait for the first response packet
119 - Do not hold the call->lock during rx_Write* and rx_Read*
120 operations when the call fields in use can only be accessed
121 by the application thread.
123 - Reduce the default window size from 128 back to 32 because
124 the current overhead from walking packet queues in the rx
125 listener thread significantly hampers the ability to
126 process incoming packets from the network.
128 - Only backoff the peer timeout once for a series of naks.
130 - Release builds are lean and mean. All debugging and
131 packet tracking code is disabled.
133 * Version number is logged to the Windows Application Event
134 Log as part of the start pending message.
136 * MSI installers now backup and restore configuration
137 data during upgrades.
139 * Rx pmtu discovery has been disabled by default. Use
140 HKLM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
141 RxPMTUDiscovery "DWORD"
142 value to re-activate.
144 * cm_BkgDaemon event evaluation has been optimized to avoid
145 unnecessary overhead.
147 * If a volume is known to be readonly, then locally fail any
148 change request without sending it to the file server. This
149 avoids triggering the file server's abort threshold.
151 * Do not leak cm_volume_t objects from the volume LRU queue.
153 * Ensure that cm_NameI errors are acted upon promptly.
154 There are many cases in the SMB server where an error from cm_NameI()
155 was either ignored or not acted upon until several other operations
156 are performed that could result in the same error being repeated.
157 This is a mistake which did not have negative side effects until
158 additional checks for callback status were added recently.
160 At present, if a CM_ERROR_ACCESS error is returned and ignored,
161 subsequent attempts to operate on the same cm_scache_t will result
162 in additional queries to the file server that will also end in an
163 abort response. This can trigger the file server to delay responses
166 * Fix fs mount point and symlink create operations which were
167 failing due to improper directory separator termination.
169 * RXAFS_GetVolumeStatus can return VNOVOL, VMOVED, etc. In
170 order to process them and update volume state a fid must
171 be passed to cm_Analyze(). Use the volume root fid.
173 * Negative caching for volume lookups that fail with either
174 VL_NOENT or VL_BADNAME. The expiration time is five minutes.
175 This prevents volume lookup storms.
177 * Provide each volume lookup its own two hour lifetime instead
178 of reseting all volume location information every two hours.
179 This prevents the first smb path evaluation after the volume
180 location data has been discarded from potentially causing an
184 * When the SYSTEM account is in use, the SMB connection is
185 not always established as S-1-5-18. Sometimes it uses an
186 anonymous, S-1-5-7, connection. In both cases the SMB
187 authentication name is the nul string. To better identify
188 when an SMB connection is local system, the smb_username_t
189 is now stored as a SID instead of a user name. When a SID
190 is used, a new flag bit is set SMB_USERNAMEFLAG_SID.
192 In smb_SetToken if AFS_PIOCTL_AFSLOGON is set and the user
193 connection is not SYSTEM, the RPC SID is checked. If RPC
194 SID is SYSTEM, that supercedes the SMB authentication.
196 * Replace STATUS_IO_TIMEOUT with RPC_NT_SERVER_TOO_BUSY for
199 * Replace STATUS_PATH_NOT_FOUND with RPC_NT_SERVER_UNAVAILABLE
200 for CM_ERROR_ALLOFFLINE and CM_ERROR_ALLBUSY.
202 * Optimize behavior of buf_CleanAsyncLocked(). Avoid searching
203 for the cm_scache_t object by FID if it is already known.
205 * If the readonly file attribute is set (stored as a unix mode)
206 then a CreateFile operation should fail if the file is opened
207 for DELETE in combination with any other privilege.
209 * If the directory buffer contents are garbage we can crash
210 the service. Add some simple validation checks to ensure
211 that cm_dirEntry_t objects have the correct flag value and
212 that the name strings are not too long.
214 * If the bulkStat errorCode indicates that a particular object
215 is inaccessible due to a VIO error, we must update the server
216 status appropriately in order to permit failover.
218 * An RX_MSGSIZE error is returned by the new PMTU detection
219 code. It is critical that such an error result in a retry of
220 the operation that failed. Otherwise, the PMTU detection can't
221 work and the server will be marked down.
222 Secondly, it is important that such errors not leak to the
223 application layer. Map them to CM_ERROR_RETRY in all cases.
225 * Translate RX_RESTARTING to the same cache manager error code
228 * When logging server volume instance errors to the windows
229 application event log, be sure to log the cell as well.
230 Translating from server ip address is non-trivial. Make it
231 easier for administrators triaging issues to plug the volume
232 and cell info into vos commands.
234 * Since unix mode bits are represented in octal in most cases
235 make sure we log them that way.
237 * The cm_enforceTrailingDot() function failed if there was a
238 trailing dot followed by white space.
240 * Rx: Do not send RX_MSGSIZE if the PMTU size did not change.
243 * A crash could occur if a single letter share name was
244 dynamically evaluated as matching a cell name. This was
245 due to a failure to treat the comparison of two nul
246 strings as identical.
248 * Docbook validation is performed by xmllint during builds.
250 * Volume package bugs in the file server can result in VOFFLINE
251 being returned to the client instead of VNOVOL or VMOVED. As
252 a result the Unix CM treats VOFFLINE the same as VMOVED and VNOVOL.
253 The Windows client has not. As a result, bugs in the file server
254 can cause the Windows client to lose if the volume has in fact
255 been moved to another server.
257 As part of this change, the volume location list is updated prior
258 to the volume status being applied to the server from which the
261 * If the pages to be flushed are from a readonly or backup volume
262 they can't be dirty. In this case there is no need to stabilize
263 the pages before they are locked and recycled.
264 Stablilization is performed on the cm_scache_t object so do not
265 stabilize and unstablize for each cm_buf_t object.
267 * Over the years the processing of the Freelance callbacks have
268 added functionality that behaves much more like FetchStatus checks
269 to a file server. If the data version of the object has changed,
270 get the new data. Given that is the case, we can remove much of
271 the original refresh logic that is rather race prone. Say goodbye
272 to cm_fakeGettingCallback and cm_fakeDirCallback.
274 * When processing a pioctl path with either smb_ParseIoctlPath or
275 smb_ParseIoctlParent, cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK|GETSTATUS)
276 must be called on the cm_scache_t object to ensure that it is up
277 to date before we permit cm_Lookup or other operations to be performed
278 on it. Add the cm_SyncOp() call to smb_ParseIoctlPath and
279 smb_ParseIoctlParent to ensure it is done for all pioctl operations.
281 * The 32-bit tools installer was displaying client configuration wizard
282 pages. Since, the 32-bit tools share configuration with the AFS
283 service, the 32-bit tools installer should not be attempting to change
284 client configuration. Remove said wizard pages from the 32-bit tools
287 * cm_data.rootSCachep is a global pointer to the cm_scache_t that
288 represents the root.afs volume root directory. Throughout the
289 code this cm_scache_t was being used without ensuring that a
290 callback to the volume is in fact valid.
292 Under most circumstances this would not be a problem. However,
293 it is possible for a request to fail due to the lack of a callback
294 at a critical moment. Add a new function cm_rootScachep() that
295 attempts to ensure that a callback is present (if possible) prior
296 to use the cm_scache_t object.
298 * When the contents of the Freelance root directory changes the fake
299 directory buffers are updated and a fakeDirVersion is incremented.
300 The dataVersion of the cm_scache_t object is supposed to be updated
301 on the next access by performing a fake get callback request.
302 Unfortunately, this did not always occur because of a race. If another
303 Freelance object is updated first, the root directory object would
304 never successfully get a fake callback.
306 This patchset ensures that the generation of the fake directory
307 buffer content and the callback are obtained under the same set
308 of locks thereby removing the race.
310 * Enforce in afsd_service that tokens can be set for alternate
311 userids only if the SetToken pioctl is called from the
312 Local System account.
315 * Revise SMB QuerySecurityInfo Response for MS10-020
317 MS10-020 (http://support.microsoft.com/kb/980232) has caused
318 many problems for implementors of SMB 1.0 servers and applications
319 that call GetFileSecurity() without checking the return code to
320 determine if the call succeeded. The gist of the vulnerability
321 was that the SMB redirector would pass any buffer it received
322 to the application regardless of whether or not it was valid.
323 MS10-020 protects the applications by strictly validating the
324 SMB response data structure and the data in the security descriptor
327 The problem for SMB 1.0 server implementors is that there have
328 been at least three different protocol descriptions for
329 NT_TRANSACT_QUERY_SECURITY_DESC published over the last decade
330 and all of them are incomplete. Therefore, just about no one but
331 Microsoft has an SMB 1.0 server implementation that produces the
332 exact out that they are expecting to validate.
334 The end result is that in an attempt to protect applications from
335 crashing due to invalid input being passed in directly caused
336 dozens of applications to crash by not returning any security
337 descriptor data at all. Even when the applications didn't crash
338 they might not have been able to save their data. Cisco WAAS
339 and NetApp DataOnTap systems were most adversely affected and
340 they have had CIFS protocol licenses for many many years.
342 To fix OpenAFS here is what needed to be done:
344 1. Instead of returning a security descriptor that gives ownership
345 to the NUL SID, give it to the Everyone SID and set the flag
346 that states that everyone has full access.
348 2. Validate the input parameters. In particular, check to ensure
349 that the SMB file descriptor is valid and the file has not
352 3. Enforce the maximum output data and parameter counts.
354 4. Handle buffer overflow and buffertoosmall conditions
355 in the manner that Microsoft expects them to be handled.
356 In particular, note that the parameter data which is returned
357 in the SMB Data Region is not counted in the Data Count.
358 Even if MaxData is 0, we can still return parameters values
359 as long as MaxParm is large enough.
361 * Prevent use of AFSCache file contents if mapped to
364 * The Windows version of "fs newcell" did not accept any parameters
365 and behaved quite differently from the Unix version. Instead of
366 permitting new cell information to be added, the Windows version
367 simply forced the existing cell information to be reacquired.
369 This update adds a new pioctl, VIOCNEWCELL2, to support the
370 implementation of a Unix-style "fs newcell". The functionality
371 added here differs from the Unix version in the following ways:
373 1. "fs newcell" with no arguments is still accepted
374 in order to maintain compatibility with prior Windows
377 2. "fs newcell -cell <cell> -dns" instructs the cache manager
378 to add the new cell but obtain the vldb server info from
381 3. "fs newcell -cell <cell> ... -registry" instructs the cache
382 manager to add the new cell and also save the cell configuration
383 data in the registry for use the next time the service restarts.
385 4. The -vlport and -fsport options are accepted although the
386 -fsport value is currently unsupported by the cache manager.
388 * New registry value "FreelanceImportCellServDB" instructs Freelance
389 to create a mount point for every cell name listed within the
392 * Path MTU discovery for Rx is activated.
394 * Rx socket input buffer is converted to a circular buffer.
396 * Fix usage of cm_FreeServerList(). Do not set the server list
397 pointer to NULL after calling cm_FreeServerList(). Doing so
398 can result in a memory leak.
400 * Only enable Rx NAT pings on a single anonymous connection at a
403 * Fix cm_IoctlSkipQueryOptions() buffer management. Prevents a
404 potential read beyond end of memory buffer.
406 * Reduce requested privileges when reading registry CellServDB
407 to the minimum required.
409 * Add support for RPC Pipe Service NetWkstaGetInfo levels
410 101 and 102 which are called on Windows 7 and 2008-R2.
412 * Prevent integer overflow during quota percent used calculation
413 in Explorer Shell Extension (RT 126846)
415 * Generate a meaningful error if "fs listacls" or "fs setacls"
416 are executed on the Freelance root.afs volume.
418 * RXAFS_InlineBulkStat errors must be processed via cm_Analyze.
419 RXAFS_InlineBulkStatus does not return errors such as EACCES,
420 VNOVOL, VNOVNODE, VOFFLINE, VBUSY, VIO, VMOVED, etc. as an RPC return
421 code. Instead they are returned in the status info errorCode field
424 Traditionally, the error associated with the first FID in the query
425 list has been returned to the caller of cm_TryBulkStatRPC().
426 However, the error has never been processed through cm_Analyze()
427 which means that the per-vnode processing for VNOVNODE and the volume
428 global processing for VMOVED, VNOVOL, etc. has never been performed.
429 As a result, failover to other .readonly volume instances cannot occur,
430 volume moves will not be handled, and files that have been deleted
433 This patchset makes the following changes:
435 1. If an inline bulk operation has been performed and the inline
436 errorCode is a volume global error, then that error replaces
437 the RPC return code within the cm_Analyze() processing for
438 the RPC. This will affect whether or not a retry operation
441 2. The variable 'inlinebulk' is reset to 0 at the top of the
442 cm_Analyze() loop in case failover from an inlinebulk capable
443 file to an inlinebulk incapable file server takes place.
445 3. The FID that is passed into cm_Analyze() is not a real fid.
446 Instead it consists of the cell and volume but vnode = 0.
447 This ensures that the error (if any) is not applied to the
450 4. If an inline bulk operation was performed, prior to performing
451 the cm_MergeStatus() operation a vnode a check is made to
452 determine if an error was returned for that vnode. If so,
453 cm_Analyze() is called with no connection, a fake cm_req_t,
454 the fid, and the error. This permits cm_Analyze() processing
455 to be performed on the file.
457 * Show configuration pages for all types of MSI installations
459 The OpenAFS MSI installer wizard used to not show any configuration
460 pages for "Typical" and "Complete" installations. Setting the
461 workstation cell and logon options during installation required
462 selecting the "Custom" option. Many users choose the "Typical" option
463 during installation, and thus would never see the configuration pages.
464 Therefore, for these users, the workstation cell was being set to the
467 This patch makes the workstation cell and logon option configuration
468 pages visible to all types of installations (except silent
469 installations which show no UI).
471 * cm_LookupInternal creates Freelance mount points and symlinks
472 when queries cannot be found in the Freelance root.afs directory.
473 If the search name is a full cell name for which vldb information
474 can be obtained, then a mount point is added. If the search name
475 is a left-most substring or the full cell name with a dot appended
476 to it, then a symlink was created. This approach created a very
477 poluted Freelance name space.
479 This patchset makes the following changes:
481 1. Do not create symlinks with a dot appended to the cellname
483 2. Do not create symlinks where the left-most substring is not
484 a full dot separated component of the cellname.
486 3. Permit lookups to succeed when we would have created a
487 symlink in the past without creating the symlink.
489 * BPlus tree lookups are much faster than searching through
490 the native directory format on Windows because the case sensitive
491 hash tables cannot be used successfully. Permit BPlus trees
492 to be used except when called with cm_BPlusDirFoo as the action
493 function because cm_BPlusDirFoo is used to build the BPlus trees
494 from the native directory format.
496 * Symlinks are ending up in the Freelance root.afs directory that
497 end with a dot. Make sure it cannot happen.
499 * cm_FreelanceAddMount and cm_FreelanceAddSymlink is supposed to
500 return the allocated FID of the entry that was added. However,
501 cm_NameI is called to perform the lookup without forcing an update
502 of the Freelance fake directory. As a result the entry may not be
505 Force an update prior to calling cm_NameI() by using
506 cm_clearLocalMountPointChange() and cm_reInitLocalMountPoints()
509 * The Freelance fake root directory buffers were not zero-filled.
510 This results in random behavior that can cause the service to
511 terminate unexpectedly.
513 * The validation check for the response from the GetVolumeStatus
514 pioctl is incorrect. The response is not simply a VolumeStatus
515 structure but also several C strings appended to it.
517 * When flushing a file, we need to commit the file length changes
518 as well as the dirty buffers. Call cm_FSync instead of buf_CleanVnode
519 which is called by cm_FSync.
521 * Prevent rx_rpc_stats global lock from being a bottleneck in the
526 * Avoid a race when updating cell vldb server lists
527 that can result in a crash.
529 * Avoid a deadlock when managing CM_SCACHESYNC_STOREDATA
530 state operations for directory objects.
532 * Add new Windows Application Event log messages for
533 VBUSY, VRESTARTING, ALL_BUSY, ALL_OFFLINE, and ALL_DOWN.
534 Include message throttling to prevent the same message
535 from being logged repeatedly within a five second window.
537 * Reduce lock contention by waiting for cm_buf_t I/O
538 operations to complete before permitting cm_SetupStoreBIOD
539 to analyze a buffer for inclusion in a BIOD.
541 * Split the cm_buf_t flags field to separate the flags
542 that are protected by the cm_buf_t mutex from those
543 protected by the buf_globalLock. This eliminates the need
544 to hold both locks everytime the flags field is accessed.
545 Both locks were not held in the past resulting in race
546 conditions that could result in deadlocks.
548 * Add "vos setaddrs" command.
550 * Rx library lock contention avoidance between rx_NewCall and
553 * Rx library races due to inconsistent use of rx_connection
554 conn_data_lock to protect the flags field.
556 * Rx library inconsistent use of RX_CALL_TQ_WAIT which could
559 * Rx library must signal transmit queue waiters when flushing.
560 Otherwise, deadlocks can occur.
562 * In cm_UpdateVolumeLocation, avoid searching for a ".readonly"
563 volume on a numeric volume name.
565 * File buffer allocations whose offsets are beyond server EOF
566 should be locally allocated and zero filled. The file server
567 should not be issued a FetchData rpc which is guaranteed to
570 * Enable integrated logon to work with Windows 7/2008 when
571 user logons are performed with a non-Domain Kerberos principal.
573 * Add Protection Error messages to aklog output.
577 * Prevent the Explorer Shell extension from crashing if
578 symlink creation failed. (126406)
580 * A Rx level NAT ping has been implemented.
581 Add NatPingInterval registry value to
582 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
583 to permit Nat Ping to be enabled. The default value is 0 seconds.
585 * When a re-initialization is taking place, be sure to reset
586 cm_noLocalMountPoints to 0 in case someone deletes the "Freelance"
587 registry key out from underneath the service.
589 * Add krb5 error message translation to aklog, afscreds,
590 afslogon.dll, the network identity manager afs provider
593 * Mode bits aren't directly exposed by the Win32 API. We were leaving
594 them to default to 0777 when creating new files and directories.
595 This version introduces two configuration parameters;
596 'UnixModeFileDefault' and 'UnixModeDirDefault' which are DWORD
597 registry entries that are used to set the initial mode bits.
598 If the values are set to 0, then the behavior is identical to what we
601 * Minidump files are now produced with a timestamp appended
604 * An SMB request debugging monitor has been added. When activated
605 the monitor will automatically turn on trace logging if any SMB
606 request has required longer than 60 seconds to complete and will
607 then create a minidump every 60 seconds thereafter until the
612 * Restore use of DNS AFSDB and SRV records by kaserver clients.
616 * Avoid a potential Freelance deadlock during initial execution
617 of afsd_service.exe if the old ini file data has to be
620 * Three rx library corrections. (1) Idle data connection
621 processing could timeout if the send window filled and
622 took longer than the idle data timeout period for the
623 transmit window to re-open. (2) The transmit queue
624 could be emptied prematurely. A required check for the
625 queue being in use was forgotten. (3) The function that
626 is supposed to implement a wait for the transmit queue
627 to cease being busy failed to wait.
630 * Restore use of DNS AFSDB and SRV records which were
631 unintentionally disabled in 1.5.69
634 * Add a context menu to the NetIdMgr AFS Provider
637 * Prevent an empty directory Btree from being created and
638 marked as valid if cm_BPlusDirBuildTree fails.
640 * cm_BPlusEnumAlloc should not fail if the enumeration
641 contains zero entries.
643 * In cm_BPlusXXX functions, return ENOMEM if malloc() fails.
645 * Do not leak the cm_scache_t dirlock if cm_BPlusDirBuildTree
648 * fs examine should report owner and group ids as signed values.
650 * Protect buffers in smb_WriteData from simultaneous writes.
652 * Query network interface MTU values from the registry because
653 the IP Helper GetAdapterAddresses() fails to report a manually
656 * Improve performance by not dropping and reacquiring the rx_call
657 lock in rx_WriteProc* and rx_ReadProc*.
659 * When configuring the rx library, the network interface MTU
660 must be applied to both the maximum send size as well as the
661 maximum receive size as is done when rx_SetMaxMTU is called
664 There was no 1.5.67 release
667 * Convert afsauthent.dll to use the xdr_alloc and xdr_free
668 routines from afsrpc.dll in place of malloc and free.
669 This permits the two libraries to be built with different
670 C runtime libraries without crashing.
672 * Set the DOS Readonly attribute on a file/directory
673 whenever the unix mode combined with the mask 0200
674 is true. Previously there was a discrepency between
675 the mask used for testing for readonly behavior and
676 that used for setting the attribute.
678 * Disable AFSVolSync based .readonly "whole-volume callback"
679 support because the all file servers prior to 1.5.67
680 (and perhaps 1.4.12) do not properly assign a value to
681 the AFSVolSync structure in bulk status RPC responses.
683 * Improve the error output from aklog to output the value
684 from krb5 error_message() if the afs_com_err output
685 indicates an unknown value.
687 * Hold a lock on the cm_scache_t object when the smb
688 server is assigning a new clientModTime value.
690 * Permit custom build version numbers to be set
691 without modifying the src/config/NTMakefile.<platform>
694 * Add support for binary code signing with cross-signed
695 certificates and arbitrary versions of signtool.exe.
697 * Convert VBUSY and VRESTARTING to CM_ERROR_ALLBUSY and
698 do not permit them to be exposed to the smb redirector.
700 * Convert STATUS_TIMEOUT responses to STATUS_IO_TIMEOUT
701 to avoid confusion within the smb redirector.
703 * Fix the byte order assigned to port numbers associated
704 with AFSDB record lookups. They must be network byte
705 order not host byte order.
707 * Add dynamic server ranking based on RPC round trip
711 * Official support for Windows 7 and Server 2008 R2
713 * Improved Rx RTT computation following the lessons
716 * Prevent a file server bug (FetchData returning an
717 invalid length instead of zero) from causing an
718 "unexpected network error" when writing to files.
720 * Promote DNS SRV records as superior to DNS AFSDB
721 records. Support arbitrary port numbers for vldb
724 * Add "fs listacl -cmd" support.
726 * Add AFSVolSync based .readonly "whole-volume callback"
727 support. With this functionality, multiple objects from
728 a .readonly volume can have their status validated by
729 issuing a single RXAFS_FetchStatus RPC.
731 * Remove drive mapping functionality and service start/stop
734 * Remove drive mapping functionality from afs_config.exe.
736 * Adjust SMB error return codes to avoid returning
737 STATUS_TIMEOUT which results in the SMB redirector
740 * Network Identity Manager OpenAFS Provider now provides its
741 own "AFS lock" notification icon to report the status of
742 "have tokens, have no tokens, service not started, service
743 started but inaccessible". Hovering over the icon lists the
744 cells for which tokens exist (if any) and the OpenAFS version
745 number. Double-clicking executes the Network Identity Manager
748 * When merging the current status for an AFS object, ensure
749 that the volume status for that object is set to 'vl_online'.
751 * Permit cm_Analyze to request a retry even when the cm_req_t is
752 flagged CM_REQ_NORETRY if the RPC failed due to RX_CALL_DEAD.
753 The retry will force the use of a new connection.
755 * Add digital signatures to resource only dlls as required by
758 * Prevent pioctl calls from retrying indefinitely when a sharing
759 violation error occurs.
761 * Add {HKLM,HKCU}\SOFTWARE\OpenAFS\Client DWORD "ShowMountTab"
762 to restore access to drive mapping functionality in afscreds.exe
765 * Permit Rx minimum peer timeout, maximum receive window size and
766 maximum send window size to be configured via the registry.
768 * The MSI installer now sets the ARPINSTALLLOCATION as required by
771 * DNS lookups set the port numbers in network byte order within
774 * Replace use of the 16-bit compatibility API WinExec with
775 ShellExecuteEx in afs_config.exe and afscreds.exe. WinExec
776 is incompatible with Windows 7 / Server 2008 R2 UAC
780 * Short circuit background volume checks if a shutdown
781 or suspend request is received.
783 * Avoid contacting the file server in order to allocate
784 an empty buffer beyond the current length of the file.
786 * When a request for streams on a directory or mount
787 point object is received, do not offer a default stream.
788 This was affecting the saving of roaming profiles.
791 * Fix Windows 2000 compatibility. Do not call rand_s()
792 even if it is compatible with the C run time library
795 * Fix a data consistency error between the output of
796 NetWkstaGetInfo and NetServerGetInfo RPCs, specify the Lan
797 workstation group name "AFS", and report server name as
798 "AFS" instead of "\\AFS" when the caller asks for "\\AFS".
800 * Enable executables to be run from \\AFS on Windows 7.
801 Return "Name not found" instead of "File not found" when
802 a directory or file name cannot be found.
804 * Prevent cache manager from marking file server "down" when
805 the data returned in response to either RXAFS_FetchData64
806 or RXAFS_StoreData64 is invalid.
808 * Add pioctl data validation to the AFS Explorer Shell extension.
811 * After 1.5.62 afslogon.dll would no longer crash but it
812 also would not load "domain" specific configurations.
815 * An uninitialized variable was present in the symlink
816 recursion detection loop.
818 * Reverse addition of unique per file GUID during
819 Create/Open responses. Inclusion of the GUID prevents
820 Cygwin applications (and others) from accessing //afs.
822 * Treat "filename::$DATA" as a synonym for "filename".
823 This addresses the recent report of roaming profiles
824 failing to save back to the file server.
826 * When given a choice between an error from rx_Write()
827 or RXAFS_EndStoreData(), use the rx_Write() error.
828 This ensures that over quota and access denied errors
831 * If an error occurs during RXAFS_StoreData operations,
832 apply that error to all of the buffers in the BIOD.
833 This ensures that failures will not be retried if they
834 will only fail again.
836 * Do not perform a DNS lookup for a cellname if the
837 cellname prefix is _._AFS_IOCTL_._. This will avoid
838 DNS lookups when the query is for
839 _._AFS_IOCTL_._.F7E5F580200909010061TTTTNT7TT.{10E39A49-4531-4496-A08E-842D4C440D20}.
841 * Fix Freelance root.afs volume object callback processing.
842 Object synchronization and status merging bugs.
844 * Fix Explorer Shell "invalid parameter" bug introduced
847 * Log RXKAD errors to the trace log.
849 * Prevent simultaneous pioctl calls from multiple processes
850 from stepping on each other's data.
852 * Prevent simultaneous pioctl calls from crashing the
855 * Add data validation to pioctl output processing in
856 ktc_GetToken and ktc_ListTokens.
858 * Add data validation to pioctl output processing in
859 fs.exe and symlink.exe.
862 * IMPORTANT: New support for DCE RPC Services: SRVSVC and WKSSVC.
863 No longer will browsing \\AFS produce truncated share names.
865 * Registry specified server preferences did not result in
866 the CM_SERVERFLAG_PREF_SET flag being applied to the
869 * Avoid unnecessary DNS lookups of share names as cells
870 when it is known that the name cannot be a cell name.
871 Any name that does not contain a dot is skipped.
873 * When processing Dfs Referral requests, do not return an
874 error if the server is down or busy; if the volume is
875 offline; the cell vldb cannot be reached; etc.
876 These paths are still \\AFS paths and so the client
877 should not be sent to search elsewhere for resolution.
879 * CRITICAL: Prevent the cm_Daemon thread from terminating when
880 the machine enters the suspend state. IF the cm_Daemon thread
881 dies, there is nothing to execute down server checks.
883 * Prevent use of smb_StartedLock before initialization which
884 can result in a panic in the lock package.
886 * Improve error checking in afslogon.dll in order to prevent
887 NULL pointer dereferences within GetLogonDomainOptions()
888 if the Lsa operations fail.
890 * Add support to the build system to automatically update
891 a Microsoft Symbol Store.
893 * Do not access the cm_conn_t after executing the cm_Analyze
894 loops. Doing so can result in an access to memory that has
897 * CRITICAL: If a StoreData request offset is not aligned on the
898 buffer module blockSize, (offset % blockSize) bytes of dirty
899 data will not be written to the file server.
901 * CRITICAL: If a StoreData64 request is sent to a file server
902 that does not support large files (>2GB) and the client is
903 unaware that the server has no large file support, the first
904 chunksize worth of data will not be written to the file server.
905 This bug was introduced in 1.5.3. It affects all IBM AFS file
906 servers and all OpenAFS file servers older than 1.4.0.
909 * If a file server becomes inaccessible while the cache manager has
910 dirty buffers to write, the afsd_service buf_IncrSync thread can
911 attempt to use 100% of the cpu.
913 * Fix the locking that protects the cm_cell_t lists
914 (all, name, id, free).
916 * Remove cm_cell_t objects from the name and id hash tables before
917 placing them on the free list.
919 * Fix "fs newcell" which was broken in 1.5.60.
921 * Do not attempt to synchronize dirty buffers if the associated
922 volume is known to be unavailable.
924 * Log to the Windows Event Log dirty buffers that cannot be written
925 during the shutdown process.
927 * Modify behavior of a Freelance mountpoint target that does not
928 specify a cell. Instead of assuming the target volume is in the
929 Freelance.Local cell, use the workstation "Cell" specified in the
930 registry. A mountpoint target of "#root.cell." will now mean the
931 root.cell volume in the workstation cell for the current session.
932 If the workstation cell changes from "athena.mit.edu" to
933 "andrew.cmu.edu", the referenced volume will also change without
934 requiring that the mount point targets be altered.
936 * Avoid false infinite loop errors when validating the cm_cell_t
937 free list during startup.
939 * Fix two more locations where xdr_free() should be used instead of
940 free() to permit a checked built of afsd_service.exe to work with
941 release builds of afsrpc.dll.
943 * Add cm_FindServerByUuid(). Re-implement RXAFS_InitCallBackState3()
944 to permit the server Uuid to be used to lookup the server object
945 and from that determine the cell. This permits callbacks that are
946 received from alternate addresses to be processed with a known server
947 object. Previously a request from an unknown server would clear all
948 callbacks from all cells.
950 * Fix a bug that prevented optimal performance when using a non-zero
951 value for 'daemonCheckVolCBInterval'. As a reminder, when
952 "daemonCheckVolCBInterval" is set to a non-zero value, all .readonly
953 volume callbacks are automatically renewed 90 minutes before their
956 * Fix automatic ranking of vldb servers whose values are obtained from
959 * Add logging for RX CALL DEAD errors which are generated if the remote
962 * Add failover for RX CALL TIMEOUT errors when the volume is readonly
963 or the call is to a vldb server.
965 * Add registry based cell search functionality to NetIdMgr,
966 afs_config.exe, and klog.exe.
968 * afsconf_GetCellInfo() has been modified to perform gethostbyname()
969 lookups on the host names in the CellServDB instead of using the
970 specified IP addresses. This provides aklog, pts, vos, etc. the same
971 CellServDB behavior that the Windows Cache Manager uses.
973 * Completely a behind the scenes change. The snprintf() routines have
974 been completely rewritten.
976 * When updating the stat cache entry callback of a .readonly object
977 from the volume group object, update the file server reference to
978 ensure it matches the most update to date callback.
980 * Add proper support for processing callbacks from multi-homed file
981 servers. Instead of comparing servers by cm_server_t pointer,
982 compare them by UUID when the UUID is known.
984 * During a shutdown short circuit the offline volume check daemon
987 * Properly track link counts during cm_Link and cm_Unlink operations.
989 * Return the error code of RXAFS_FetchData / RXAFS_StoreData in
990 preference to an error code reported by rx_EndCall.
992 * Add "PerFileAccessCheck" registry value to permit testing against
993 experimental file servers that include per-file acl support. This
994 value is intentionally undocumented. It is not to be used by
995 production environment deployments.
997 * Log a warning to trace log if a volume group is marked as a DFS
998 File Set. The Windows OpenAFS client does not include DFS support.
1000 * Fix a bug introduced in 1.5.60 that prevents the afs netidmgr
1001 provider from obtaining tokens when referrals are in play.
1003 * Add "fs chown" and "fs chgrp" commands to permit the owner and
1004 group of objects stored in AFS to be set from Windows.
1006 * Avoid performing background daemon operations when the machine is
1007 going into suspend mode.
1009 * Perform offline volume checks in most recently used order.
1011 * Prevent crash when a data version for a cache object goes backwards.
1013 * fs setquota, fs setcachesize, vos setfields, and vos create now
1014 accept human readable orders of magnitude. (K, M, G)
1016 * fs listquota fixed to permit large quota sizes to be displayed.
1018 * rx packet resend and data packets sent counts were incorrect.
1020 * Multi-thread safe library versions are now being generated and used.
1021 mtafsubik.lib, mtafsutil.lib, mtafsvldb.lib, mtafsvol.lib.
1023 * Microsoft SMB Redirector (mrxsmb.sys) support for ExtendedSessTimeout
1024 values are now available on XP through Windows 7. Add functionality
1025 to autodetect if such support is present on the machine. If so,
1026 configure it if necessary and dynamically adjust the AFS Rx timeout
1029 * Correct another race condition in the Rx library that could result
1030 in an unexpected panic while freeing the Rx call iovq.
1033 * A fix to the pioctl library to support drive substitution
1034 to UNC paths. (SUBST <d:> <\\afs\cell\path>).
1036 * On April 9th Microsoft released a Hot Fix for Windows Server 2003 SP2
1037 that corrects a deadlock in the smb redirector and also adds new
1038 functionality that permits the AFS SMB server to be given a longer
1039 timeout than is normally the case. New functionality has been added
1040 to configure these additional LanmanWorkstation\Parameter values.
1041 (This functionality has been backported to XP SP3 and is scheduled
1042 to be released on June 5th.)
1044 * The BackConnectionHostNames registry value configuration was broken
1045 when dynamic re-establishment of Netbios Name registrations was added.
1048 * Hidden vos commands are revealed.
1050 * If the "DisableLoopbackCheck" registry value is set, do not unset
1051 it during the same service session.
1053 * Reorganize code that prevents multiple Store operations to the same
1056 * Modify IsPathInAfs test in Explorer Shell Extension and fs.exe to
1057 permit broken symlinks to be treated as being in AFS.
1059 * Fix vos commands that output 64-bit integer values.
1061 * Cygwin Import Libraries are provided in the SDK for all OpenAFS DLLs.
1062 Permits building cygwin applications against OpenAFS libraries.
1064 * OpenAFS Release Notes, Administrator Guide and User Guide now installed
1065 as Windows HTML Help (.CHM) files.
1067 * NSIS installer does a much better job of cleaning up files left
1068 over from previous installs.
1070 * Fix RT#124787, a race condition between "fs flush <dir>", "fs flushvolume",
1071 or "fs flushall" and on-going directory operations that can result in
1072 afsd_service.exe crashing.
1074 * Add support for DNS SRV records in place of AFSDB records.Â
1075 _afs3-vlserver._udp.<cellname>. Priority field is used.Â
1076 Weight is currently ignored.
1078 * Add a method of specifying Client CellServDB information within the
1079 registry that can be used to either override the CellServDB file or
1080 force the use of DNS lookups for a given cell.
1082 The registry schema is as follows:
1084 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\
1085 "LinkedCell" REG_SZ "[cellname]"
1086 "Description" REG_SZ "[comment]"
1087 "ForceDNS" DWORD {0,1}
1089 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\[servername]\
1090 "HostName" REG_SZ "[hostname]" Default: [servername]
1091 "IPv4Address" REG_SZ "[address]" Used only if gethostbyname() fails.
1092 "IPv6Address" REG_SZ "[address]" <future>
1093 "Comment" REG_SZ "[comment]"
1094 "Rank" DWORD "0..65535" Default: 0
1095 "Clone" DWORD "{0,1}" <future: server only>
1096 "vlserver" DWORD "7003" <future>
1097 "ptserver" DWORD ... <future>
1099 ForceDNS is implied non-zero if there are no [servername]
1100 keys under the [cellname] key. Otherwise, ForceDNS is zero.
1101 If [servername] keys are specified and none of them evaluate
1102 to a valid server configuration, the return code is success.
1103 This prevents failover to the CellServDB file or DNS.
1105 Only one of "HostName" or "IPv4Address" is required. "HostName"
1106 is optional if [servername] is the host name.
1108 * Extend registry based CellServDB functionality to
1109 afsconf_GetCellInfo() interface used by aklog and many
1110 other command line utilities.
1112 * libafsconf.dll moved from OpenAFS\Client\Program to OpenAFS\Common
1113 as it is now used by both client and server components.
1116 * PriorityClass of afsd_service.exe process raised to
1117 "High" to match the priority of the system services
1118 that are waiting for afsd_service.exe to respond.
1119 Configurable using [TransarcAFSDaemon\Parameters]
1120 DWORD "PriorityClass" value. (use values from Win32
1123 * Change error reporting of CM_ERROR_BADSHARENAME to
1124 use the correct error (NTSTATUS_BAD_NETWORK_PATH).
1125 The error we were sending would cause the smb
1126 redirector to drop the connection.
1128 * Change the error reporting of CM_ERROR_ALLOFFLINE
1129 and CM_ERROR_ALLDOWN to return NTSTATUS_PATH_NOT_FOUND.
1130 Previous we would return Bad Network Path which
1131 would cause the smb redirector to cancel all file
1132 handles on the same "share".
1134 * A debug version of afsd_service.exe can now be used
1135 with a release version of afsrpc.dll (or vice versa).
1136 afsrpc.dll now implements and exports xdr_free().
1138 * Removed all unprotected OutputDebugString() calls as
1139 those calls can block for up to 10 seconds if there
1140 is contention for the globally shared memory or if
1141 the debugger fails to read data in a timely manner.
1143 * Visual Studio 2008, WDK 6.1 and DDK 6.1 are now
1144 supported for builds. The production releases still
1148 * Discovered that XP/2003 supports a variant on recvmsg
1149 called WSARecvMsg. The input parameters are mostly
1150 compatible with the Posix recvmsg. This will save
1151 multiple memcpy() calls per received packet.
1153 * Discovered that Vista/2008 supports a variant on
1154 sendmsg called WSASendMsg. The input parameters
1155 are mostly compatible with the Posix sendmsg. This
1156 will save multiple memcpy() calls per sent packet.
1158 * Exporting rx_ReadProc32, rx_WriteProc32 from afsrpc.dll
1159 and utilizing rx_ReadProc32 from within the cache manager
1160 as part of the FetchData operations. rx_ReadProc32 and
1161 rx_WriteProc32 are optimized versions of rx_ReadProc and
1162 rx_WriteProc for use when reading 32-bit integer values.
1164 * The Network Identity Manager AFS provider now reports
1165 progress events. These provide a better user experience
1166 with Network Identity Manager 2.0.
1168 * [RT 124293] A race condition exists which permits the
1169 scp field of the an smb_fid_t object to become invalid
1170 after a request on the smb_fid_t has begun. Remove the
1171 race by protecting all access to the scp field with the
1172 smb_fid_t mx mutex and obtaining a local reference on
1173 the cm_scache_t object for the length of the request.
1175 * [RT 124276] If the vldb is out of sync with the contents
1176 of the file servers, afsd_service will retry too many
1177 times when a file server reports a volume as not being
1178 present. Now if the list reported by the vldb is the
1179 same as the previously seen list, then the retry is
1182 * [RT 124276] Read-only volume failover was broken in
1183 1.5.53 whenever accessing a volume results in VNOVOL or
1186 * [RT 124276] Prior to 1.3.70 the volume server reference
1187 list was not reference counted and would be prematurely
1188 freed while in use. When reference counting was added
1189 in 1.3.70 a bug was introduced that could result in
1190 service reference list corruption.
1192 * Add Windows Application Event Log warning messages for
1193 "Client SMB MPX value too large" and
1194 "Client SMB Buffer Size too small".
1196 * Renaming of files across directory boundaries would result
1197 in an invalid handle error when attempting to access the
1198 files after the move.
1200 * Fix the handling of Tran2 Set Path Info RPCs. Do not fail
1201 when a smb file descriptor cannot be found. The whole point
1202 of using a Path Info function is because an smb file
1203 descriptor wasn't allocated.
1205 * More edge cases in which dynamic addition of Freelance
1206 root.afs entries would get the wrong FID or where the
1207 root.afs directory would not be refreshed.
1209 * Buffer overflow could occur if the workstation cell name was
1210 longer than 64 characters. Crashes could occur in
1211 afscreds.exe, afslogon.dll, and afsd_service.exe.
1213 * VNOSERVICE and VOFFLINE errors were leaking and were exposed
1216 * Improve consistency of the bad data version value used for
1219 * Log file server uuid values as part of the cm_server object
1220 when available. Dump the cm_server object list in response
1223 * Optimize the performance of resetting access control lists
1224 when tokens are set or removed.
1226 * Fix vos.exe so that it can work with very large volume
1229 * Remove symlink recursion tests and increase max symlink count
1233 * Undo the "UAC manifest fix" applied to afs_config.exe.
1235 * Fixed some additional race conditions in the Rx library
1237 * Ensure that Freelance allocation of vnodes follow
1238 the AFS convention of odd vnodes are directories
1239 and everything else is an even vnode.
1241 * Add Freelance logic to mount point and symlink
1242 evaluation functions.
1244 * Enhance smb_ParseASCIIBlock() so that it can handle
1245 all of the STRING formats defined by the CIFS Technical
1248 * Validate the output of smb_ParseASCIIBlock() in all
1249 callers. Return CM_ERROR_BADSMB if the STRING field
1250 cannot be parsed. CM_ERROR_BADSMB will cause the
1251 contents of the packet to be logged.
1253 * If multiple SMB Raw Write operations were taking
1254 place at the same time, there could be data corruption
1255 because unique event objects were not generated for
1256 each Netbios receive operation.
1260 * Fix a race in rxi_WriteProc which could result
1261 in a freed or in use packet being passed to
1262 rxi_PrepareSendPacket if the call is currently
1263 in a transmit wait state. Passing a freed packet
1264 will result in a panic. Passing an in use packet
1265 might result in a call failure or perhaps data
1268 * Kerberos referrals broke the NIM AFS provider
1269 logic for automated configuration of the workstation
1270 cell with a new identity. Now if a referrals
1271 response is returned, assume the cell belongs to the
1274 * The interface to buf_CleanAsync() and
1275 buf_CleanAsyncLocked() has been changed to permit
1276 errors to be propagated to the caller.
1278 * During an enumeration of a directory, if the btree
1279 has not yet been generated, return an error.
1281 * When mapping CM_ERROR values to NTSTATUS values
1282 do not treat 0 as an error.
1284 * Avoid a cm_scache_t object reference count underflow
1285 when renaming via the SMB interface.
1287 * smb_IsMask() can be called with a non-8.3 input buffer.
1288 Separate smb_IsMask() into smb_Is8Dot3Mask() for 8.3
1289 input buffers and smb_IsMask() for non-8.3 input buffers.
1291 * Add cm_buf_t error state to the "fs memdump" output.
1293 * pioctl drive mapping checks need to take into account
1294 Global Drive Mappings and SUBST assignments.
1296 * Rename operations should not be permitted if the
1297 destination name collides with an existing name that
1298 differs only by case UNLESS the collision is the object
1299 that is being renamed.
1301 * Increase the maximum number of background daemon threads
1302 to 64. Leave the default at 4.
1304 * Set the FILE_UNICODE_ON_DISK flag when reporting the
1305 file system characteristics.
1307 * Report the maximum component name length as 255 even
1308 for Unicode as part of the file system characteristics.
1310 * Add RX debugging information to "fs memdump" output
1313 * Refcount the cm_server_t objects returned by
1316 * Fix the UAC manifest applied to afs_config.exe
1318 * rxdebug now indicates if any RX_CONN flag is set
1319 not just some of them
1321 * Change the mapping of CM_ERROR_WOULDBLOCK from
1322 NTSTATUS_LOCK_NOT_GRANTED to NTSTATUS_CANT_WAIT
1326 * Includes #GCO Public CellServDB 07 Nov 2008
1328 * Prevent separate cell entries from being created for
1329 "foobar.org" and "foobar.org." which can happen if
1330 an AFSDB record is found instead of a CellServDB
1333 * Fix a panic when checking the SMB virtual circuit list.
1334 If a VC is destroyed while the list is being checked
1335 it is possible for the next object to become invalid.
1337 * Increased logging to afsd_init.log of netbios name
1340 * Fix a DNLC panic when a UTF-16 string that cannot be
1341 converted to a utf-8 string is provided as a filename.
1343 * In the NSIS installer, delete old afslogon.dll/afscpcc.exe
1344 from %windir%\system32 and fix the WinLogon\Notify\AfsLogon
1345 location to refer directly to the new location in
1346 %programfiles%\openafs\client\program
1348 * Be more aggressive about responding to Netbios
1349 NRC_BRIDGE/NRC_NOWILD errors. Rebind the Netbios name to
1350 the LANA within the listener thread without performing a
1353 * Use the WNet Enumeration of all connected drives as a filter
1354 in the pioctl GetIoctlHandle operation. This should remove
1355 the last situations in which there was a delay when opening
1356 an Explorer Shell properties menu with the AFS Shell
1359 * Fix the parsing of CellServDB in Windows to handle linked
1360 cells as is done on Unix.
1362 * Add support for linked cells to the cache manager,
1363 netidmgr afs provider, and aklog. When two cells are
1364 linked, a volume lookup that fails in one cell is looked
1365 up in the other. This can be used for construction of
1366 test cells that override specific volumes; cell renaming;
1369 * Increase the length of cell and realm names that can
1370 be entered in the NIM AFS provider.
1372 * Actively prevent a 32-bit afsd_service.exe from
1373 executing in the WOW64 environment.
1375 * Protect against invalid IP address values in CellServDB
1380 * Fix an Rx panic caused by an attempt to double free
1381 an rx_packet when an error occurs while writing the
1384 * Fix an exception cause by a race condition when
1385 processing multiple file change notifications registered
1386 by the Explorer Shell or other applications.
1388 * Modify the Rx debug interface to use var_args and
1389 permit Rx debugging to be used on 64-bit Windows
1390 which previously had it disabled.
1394 * Automated lock hierarchy enforcement added to the
1395 osi base lock package. This permits detection of lock
1396 violations that might but often does not result in
1397 deadlocks. Enabled by default in the checked builds.
1398 May be activated in release builds via the registry
1399 value "LockOrderValidation" in the TransarcAFSDaemon
1402 When a violation is detection an assertion is thrown
1403 generating a dump file and a message indicating which
1404 locks were obtained out of order is logged to the
1405 Windows Application Event Log.
1407 * Attempt to ensure that a dump file will be generated
1408 whenever the service panics.
1410 * Removed unused lock cm_bufGetMutex
1412 * Many out of order lock acquisitions were detected
1413 using the lock order verifier and were corrected.
1415 * A race condition within the function path cm_GetSCache() ->
1416 cm_GetNewSCache() -> cm_RecycleSCache() permitted a
1417 cm_scache_t object to be issued simultaneously to two
1418 threads. This would eventually result in a panic due
1419 to the resulting undercount.
1421 * Fix interpretation of the empty string as the ioctl path
1422 to mean the current directory. "fs lsm", "symlink list",
1423 etc. now return a "not a ..." error instead of "does not
1426 * Add SMB_STRF_SRCNULTERM flag to smb_ParseStringXXX functions.
1427 to indicate that the input string is a nul terminated string.
1428 Assign ti when input strings are nul terminated instead of
1429 length counted. (from 1.5.50)
1431 * The CIFS protocol specification for handling NT_TRANSACT_CREATE
1432 does not match the observed behavor. The 'nameLength' is specified
1433 in bytes, not in characters. Fix the implementation to match
1434 the observed behavior. (1.5.50)
1436 * The cm_HaveAccessRights() attempt at deadlock avoidance by
1437 calling lock_TryRead() on the parent directory cm_scache_t rw-lock
1438 does not avoid the deadlock. Avoid the deadlock by enforcing the
1439 lock order of the lowest vnode first. Then remove the infinite
1440 loop avoidance in cm_SyncOp() that was returning an unwarranted
1441 access denied error.
1443 * File Server lock synchronization was not propertly enforced in
1444 smb_ReceiveNTCreateX() and smb_ReceiveNTTranCreate().
1446 * Another change to the short name generator. Plus and equal are
1447 not valid characters in a short name according to the
1448 RtlIsNameLegalDOS8Dot3() in kernel mode. They have been replaced
1449 with left and right brace in order to minimize the impact.
1451 * Fix a race condition that can result in a heap overwrite error if
1452 the number of known file servers or vldb servers increases during
1453 the preparation phase of a cm_CheckServers() operation.
1455 * If the SMB Authentication would fail, a LSA allocated memory
1456 buffer would be leaked.
1458 * Rewrite the cm_cell_t allocation code to permit cell vldb server
1459 lookups to fail without leaking cm_cell_t objects.
1461 * Add "fs setcrypt auth" to permit the use of rxkad authentication
1462 only mode. Not documented.
1464 * Initialize Rx mutex and lock objects once and only once as part of
1465 the DLL load process.
1467 * When reporting the Volume Name "AFS" do not include the terminating
1468 NUL in the reported length.
1470 * Refresh Volume Location information upon receipt of a VNOVOL error
1471 from the file server and avoid a race condition that prevented the
1472 updated volume location information from being destroyed immediately
1473 after it was acquired.
1475 * In response to VNOVOL and VMOVED, remove the volume id from the server's
1476 volume list so that the volume's up and down status is not affected
1477 when the machine is shutdown.
1479 * Many fixes to the Rx stack related to packet allocation and deallocation;
1480 Rx debugging; Rx packet tracking across queues; and Rx packet loss caused
1481 by overwriting the rx_call currentPacket pointer when it was already
1482 referring to a packet.
1484 * Rx Jumbograms are now off by default. To re-enable, add "RxJumbo" = "1"
1485 in the TransarcAFSDeamon/Parameters key.
1487 * "fs flush", "fs flushvolume", and "fs flushall" now destroy locally
1488 constructed directory B+ trees for flushed directory objects.
1490 * When cached pages for directory objects have been locally modified, do
1491 not permit the locally modified pages to be mixed with pages subsequently
1492 obtained from the file server. Doing so can result in directory
1495 * If Ubix errors are received in response to VL_xxx operations record
1496 the error on the server instance and permit the operation request to
1497 be retried in case another server is able to answer the query. We have
1498 seen at least one instance of a cell in which a VL server returned
1499 UNOQUORUM (possible database truncation) when the other servers were
1500 responding successfully.
1503 * Optimize the assignment of write errors to dirty
1504 buffers by applying the first error to all buffers
1505 belonging to the same object instead of trying each
1508 * Add resource version info to xstat_fs_test and xstat_cm_test
1510 * Fix a delete vs delete[] usage problem in the WiX custom
1513 * When creating a mount point, validate the volume name to
1514 make sure it exists.
1516 * When processing a mount point, pay attention to VL_BADNAME
1519 * Remove a race condition from cm_volume_t recycling that could
1520 have resulted in two threads making use of the same object at
1523 * Fix a reference count leak on smb_vc_t objects introduced in
1524 1.5.50. Permit the objects to be freed once again.
1526 * Fix the output of "fs memdump" for smb_vc_t objects that was
1529 * Add routines for debugging smb refcount errors.
1531 * When shutting down the cache manager, do not permit AFS
1532 callback requests to be processed. Doing so can result in
1533 invalid memory accesses.
1535 * Do not recycle a deleted cm_scache_t object unless the
1536 refCount is 0. Doing so can result in a deadlock and
1537 invalid memory access.
1539 * Minimize the number of RPCs sent to the file server for
1540 directory bulk status requests. Only request status for
1541 objects that we are actually reporting on and only if we
1542 do not already have status for the object and then make
1543 sure that we return as much as we can at once via the SMB
1544 interface but not so much that it exceeds the RDR timeout
1546 * Enable bos restricted operations
1548 * Fix the creation of submounts used by the afscreds and afs_config
1551 * Avoid repeated retries when attempting to obtain access rights.
1552 If the access rights were obtained successfully but do not
1553 satisfy the request, fail the request.
1555 * Fix error code reported when attempting delete a file on a
1556 readonly volume or one that is marked with the readonly DOS
1559 * Fix a short name truncation bug.
1561 * Fix a heap corruption error when reading the CellServDB file
1564 * Fix directory listings when the directory has "lookup" permission
1565 only and the directory has never been previously accessed.
1567 * Fix several tracking issues for cm_scache_t rw locks.
1569 * Ensure that all buffers used to store cell names are the
1572 * Restore Windows 2000 support (broken in 1.5.50).
1573 LOCALE_INVARIANT was introduced in XP.
1575 * Add RxUdpBufSize registry value. Default is 256K.
1577 * Avoid multiple deadlocks between cm_freelanceLock and cm_scacheLock
1579 * Do not include trailing NULs when returning file names in
1580 a directory search response.
1582 * Prevent RX from crashing the afsd_service by attempting
1583 to allocate to many buffers at once when memory is fragmented.
1585 * Avoid a deadlock when freeing cm_server_t objects
1587 * Optimize search cell file. Do not search to end of file
1588 if the cell has already been found.
1590 * Remove explorer shell extension delay on local disks.
1592 * Automatically pre-allocate 64 Rx packets per thread
1594 * Add smb lock requests and stat cache lock allocations to
1595 the output from fs memdump.
1598 * Fix Large File Support (broken in 1.5.50)
1600 * Fix "fs setacl" (broken in 1.5.50)
1602 * Fix pioctl path evaluation of drive letter
1603 mapped paths. (broken in 1.5.50)
1605 * Fix installation of VC8 Runtime library in the EXE
1606 Retail installer. (broken in 1.5.50)
1608 * Avoid a potential deadlock during Volume Location
1609 updates if all the file servers are unreachable at
1610 the time the VL update succeeds.
1612 * Ensure that rx connection object is references
1615 * Fix propagation of over quota errors during cm_FSync()
1618 Since 1.5.39 [1.5.50 released 16 July 2008]
1619 * There were no releases between 1.5.39 and 1.5.50.
1621 * FIRST RELEASE WITH UNICODE SUPPORT
1623 * Pioctl interfaces to the cache manager refactored
1624 to provide layering between the SMB specific code
1625 and the general purpose ioctl operation.
1627 * Garbage collect dead SMB virtual circuits as soon
1628 as they are no longer being referenced. This avoids
1629 problems with outstanding locks not being dropped
1630 when the virtual circuit becomes invalid.
1632 * Remove the IBM Administration Reference documentation
1633 and replace it with the OpenAFS Command Reference
1636 * Avoid calling rx_SetDeadTime and rx_SetHardDeadTime
1637 functions each time a connection is about to be used.
1638 Do not hold a lock on the rx connection object while
1639 it is being selected. This avoids a race between
1640 threads attempting to set the timeout values and
1641 removes a bottleneck that was hampering performance.
1643 * Ensure that the smb directory attribute is set
1644 for all directory objects.
1646 * Add vs2008 support to the NSIS installer scripts
1648 * Replace the VC Runtime EXE installer with the MSI
1649 installer in the NSIS installer scripts
1651 * Properly delete the VC7.1 C Runtime libraries.
1653 Since 1.5.36 [1.5.39 released 23 June 2008]
1654 * There were no 1.5.37 or 1.5.38 releases for Windows
1656 * Use Visual Studio version of vsprintf() instead of
1657 home grown version that resulted in linking conflicts
1659 * Prevent buserver, ptserver and fileserver from crashing
1660 due to non-portable strftime() format strings.
1662 * Prevent server crashes caused by rx_InitMorePackets()
1663 being called prior to allocation of all of the required
1664 synchronization objects.
1666 * Fixed a memory leak in the internal B+ tree directory.
1667 For objects with non-8.3 names, the 'longname' memory
1668 allocation would be leaked during object deletion or
1671 * Fixed a memory leak in the background daemon cm_CheckServer()
1672 routine. The memory allocated to store the server list
1675 * Obtain a missing lock around a call to cm_RemoveSCacheFromHashTable().
1677 * Correct an abstraction layer violation. cm_scache_t objects
1678 should be marked deleted in cm_Unlink() and cm_RemoveDir() and not
1679 in smb_CloseFID(). Cleanup of deleted cm_scache_t objects should be
1680 performed in cm_ReleaseSCache() when the reference count hits zero.
1682 * Prototype cm_AdjustScacheLRU() and re-implement it using osi_QAddH().
1684 * Do not reference 'smb_logp' in cm_vnodeops.c. Use 'afsd_logp'
1687 * Ensure that scache object 'nextp' pointers are set to NULL
1688 when the objects are removed from the hash table linked list.
1690 * If a dirty buffer cannot be written to the file server because
1691 of an access denied error, mark the error in the cm_buf_t
1692 structure so that it can be dealt with instead of retrying
1696 * Update CellServDB file to GCO Public 23 Apr 2008
1698 * Fix a cm_buf_t reference count leak when attempts to write
1699 dirty buffers to the file server from within cm_IncrSyncer()
1702 * Prevent udebug from crashing.
1704 * Activate RX Idle Timeouts. If the file server is busy for more
1705 then 30 seconds, attempt to failover to another server without
1706 marking the busy server down.
1708 * Another VNOVNODE issue fixed. When writing a dirty buffer
1709 to the file server, if VNOVNODE is received, mark all buffers
1710 as invalid without further attempts to contact the file server.
1712 * Improved performance on high latency links.
1715 * Fix a bug preventing the re-initialization of the Freelance
1716 root.cell contents when a change is made.
1718 * Properly return CM_ERROR_NOSUCHFILE when removing a symlink
1719 or mountpoint that does not exist.
1721 * aklog, afscreds, integrated logon, and the NIM AFS credential
1722 provider will all attempt to first acquire an afs service ticket
1723 using the service principal: afs/<cell>@<CLIENT_REALM>. This
1724 permits the proper detection of the cell realm when multiple
1725 realms are local realms for the cell.
1727 * Fix fs commands that re-use file identifiers instead of forcing
1728 a path evaluation each time. This was broken in 1.5.33 when
1729 hashes were added to the cm_fid_t structure.
1731 * Fix symlink and mountpoint removal and then recreation when
1732 applied to the fake Freelance root.afs volume.
1734 * Fix SMB change notifications for the Freelance root.afs volume
1735 when creating or removing symlinks and mount points.
1737 * Force DNS AFSDB cell name lookups to lower case if the DNS
1738 response is not all lower case.
1740 * 32-bit installers are now built with Visual Studio 2005
1742 * All vos.exe commands now support the "-noresolve" option which
1743 forces the display of IP addresses instead of DNS hostnames.
1745 * No longer hold the cm_volume_t mutex across VL_xxx RPCs. This
1746 permits operations that do not require up to date volume
1747 location data to continue while waiting for the RPC to complete.
1748 Multiple threads waiting for a volume location update on the
1749 same volume set no longer trigger multiple calls VL_xxx RPCs.
1751 * Volume server references could be marked offline and never
1752 reset to the not_busy state.
1754 * Further optimizations to cm_Analyze() retry logic for
1755 CM_ERROR_OFFLINE and CM_ERROR_ALLBUSY error handling.
1758 * Optimize the DNLC by applying Interlocked incrementing for
1759 statistics gathering and enabling greater use of read
1760 locks instead of write locks
1762 * Further optimize osi_Log operations when logging is disabled
1763 by removing the need for the function call overhead.
1765 * Add a new lock operation, Convert Read to Write, and apply
1766 it throughout the cache manager. This function provides a
1767 fast transition from a read lock to a write lock when the
1768 caller is the only reader.
1770 * Fix a timeout problem when opening files or request locks
1771 when the file in question is already write-locked by another
1774 * Fix a deadlock that can occur if "fs flushall" is executed
1775 while applications are writing to AFS.
1777 * The cache manager would re-read data from the file server
1778 after file truncation when obtaining buffers to write to the
1779 same file until the first StoreData RPC completes.
1781 * Fixes error handling in cm_NTCheckDelete()
1783 * Replaces cm_scache_t mutex with a read-write lock permitting
1784 additional parallel access.
1786 * Prevent an cm_scache_t reference undercount when evaluating
1787 symlinks containing @sys when none of the active sysnames
1788 match any of the available targets.
1790 * Prevent the leak of a cm_scache_t rwlock if associated symlink
1793 * Set RxMaxMTU registry value to 0 which is the equivalent of
1794 it not being set at all.
1796 * Reduce contention between BkgDaemon threads
1798 * Prevent CM_SCACHEFLAG_ASYNCSTORE from being reset on a write
1801 * Convert reference counts on cm_volume_t, cm_conn_t, cm_server_t
1802 objects to use the faster Interlocked operations which avoid
1803 the need to hold write locks.
1805 * Permit rxkad errors other than RXKADEXPIRED to be treated as a
1806 non-fatal error that forces a retry against another service
1809 * Fix a crash in smb_WriteData that would occur if an application
1810 opened the file as read-only and then sent a write request anyway.
1813 * Respond to VL_NOENT errors by removing cm_volume_t object and
1814 setting it to be the first object to be recycled. This is not
1815 only a more efficient use of resources but it also prevents
1816 repeated attempts to query the VLDB server for the volume from
1817 the CheckOfflineVolumes routine.
1819 * Change the defaults for RxEnablePeerStats and RxEnableProcessStats
1822 * A first cut at a cache manager statistics monitor that can be used
1823 to determine the necessary cache parameters to support the
1824 application working set.
1826 Off by default, the performance package can be activated by setting
1827 "daemonPerformanceTuningInterval" in the service Parameters key.
1828 As with the other daemon interval values the unit is in seconds.
1830 At service start and each succeeding interval the cache manager
1831 will write statistics to %TEMP%\afsd_performance.txt showing the
1832 relative usage of cm_scache_t, cm_volume_t and cm_buf_t objects.
1833 The FID statistics keep track of all FIDs seen by the cache manager
1834 during the service session whether or not they are backed by any
1835 live object in the cache.
1837 These statistics are not stored in the cache file.
1839 * Fix windows event logging broken in 1.5.33.
1841 * Eliminate multiple calls to syscfg_GetIFInfo() which is used to
1842 populate the IP address list for WhoAreYou/TellMeAboutYourself RPCs
1843 as well as when computing the server preference list.
1845 * When merging status after a StoreData operation prevent data
1846 buffers from being discarded that we would prefer to keep.
1847 (Bug introduced in 1.5.33.)
1849 * Modify installers the set RxMaxMTU to 0.
1851 * In the AFS Configuration Panel on the Advanced->Miscellaneous
1852 dialog, the Background Daemons and Service Threads fields were
1855 * Increase default number of Background Daemons to 4.
1857 * Remove cm_volume_t reference from the cm_scache_t object. Permit
1858 cm_volume_t objects to be recycled.
1860 * Windows Error Reporting has shown a number of previously unseen
1861 stack corruption errors in the Explorer Shell extension. The
1862 minidump appeared to indicate the problem was in ParseAcl().
1863 Reviewing the function it appears that numerous things could go
1864 wrong if invalid input was provided. Added error checking and
1865 data validation. Hopefully this will address the problem.
1867 * Windows Error Reporting has shown crashes in aklog as a result
1868 of krb5_cc_get_default() failures not being handled. They are
1872 * The Rx library used a 32-bit type for sockets which was
1873 truncating the socket value on 64-bit Windows. This
1874 was not an issue in general but became an issue when
1875 Rx "hot thread" support was enabled.
1877 * Enable "Rx hot thread" support which ensures that while
1878 a received message is being processed that a new thread
1879 is allocated to receive the next message to arrive.
1881 * Increased parallel access to smb_rctLock protected
1882 resources. Reduce the number of times that the
1883 lock is dropped and re-obtained.
1885 * Increased parallel access to cm_scache_t and cm_buf_t
1886 resources by replacing the previous global write lock
1887 protected reference counting mechanism with one built
1888 upon the use of InterlockedIncrement and InterlockedDecrement.
1890 * Implement a hash for cm_fid_t to reduce comparison costs
1891 and inline the cm_FidCmp() function.
1893 * Redefine the BUF_HASH and BUF_FILEHASH in terms of the cm_fid_t
1894 hash which has a better distribution
1896 * Modify cm_ConsiderPrefetch to evaluate the amount of data in
1897 the most recent read request instead of the next chunkSize.
1898 cm_chunkSize can be dozens or hundreds of buffers. As a result
1899 too much time is spent performing the evaluation.
1901 * Fix the usage of cm_scache_t bufCreateLock. The purpose of this
1902 lock is to prevent the creation of new buffers while a truncation
1903 is being performed. All references to bufCreateLock have been
1904 removed except in two places:
1905 i. a write-lock surrounding the function that calls buf_Truncate()
1906 ii. a read-lock within buf_GetNewLocked() that actually allocates
1909 * Modify the CM_CONFIG_DATA_MAGIC value to include a
1910 CM_CONFIG_DATA_VERSION number which value be used to force the
1911 replacement of the cache file contents when incompatible changes
1912 are made between releases.
1914 * CM_SCACHESYNC_ASYNCSTORE should not be ordered by
1915 cm_SyncOpCheckContinue
1917 * Avoid calls to multi_Rx if nconns == 0
1919 * Modify smb_WriteData to perform background writes based upon the
1920 crossing of 'smb_AsyncStoreSize' boundaries instead of cm_chunkSize
1921 boundaries. This will slow down writes from the SMB interface but
1922 will avoid the risk of the CIFS client disconnecting from the AFS
1923 client SMB server. The default value for 'smb_AsyncStoreSize'
1926 * SMB AsyncStore functionality is on by default but can be
1927 disabled using the registry value:
1929 HKLM\SOFTWARE\OpenAFS\Client DWORD "EnableSMBAsyncStore"
1931 smb_AsyncStoreSize can be adjusted with
1933 HKLM\SOFTWARE\OpenAFS\Client DWORD "SMBAsyncStoreSize"
1935 The value must be a multiple of the buffer block size
1936 and cannot be larger than the chunk size.
1938 * Do not leak a read lock on the bufCreateLock
1940 * Do not forget to call cm_SyncOpDone if cm_SyncOp succeeded
1943 * In MergeStatus, only remove clean buffers from the hash table.
1945 * Avoid the need to update the data version number on each buffer
1946 associated with a scache when MergeStatus is called after a
1947 StoreData by maintaining a range of valid data versions as part
1948 of the cm_scache_t object.
1950 * Profiling shows large numbers of blocked calls in cm_HoldSCache
1951 from cm_HaveAccess... We can safely avoid the call so do so.
1953 * Add "ForceAnonVLDB" service parameters registry value. Default
1954 is 0. Set to 1 to force the use of anonymous connections to
1955 the VLDB servers. This is useful when using krb5 tokens and the
1956 VLDB servers have not been updated.
1959 * Do not attempt to search for a cell name that matches one
1960 of the special share names "svrsvc", "wkssvc", "ipc$"
1962 * Issue permission denied warnings from afscreds if attempts
1963 to modify the startup shortcut or registry values fail.
1964 This happens under Vista UAC.
1966 * Do not return BAD_NETWORK_ERROR in response to attempts
1967 to create directories or files in the Freelance root.afs
1968 volume. Instead return ACCESS_DENIED.
1970 BAD_NETWORK_ERROR was being sent because the Freelance cell
1971 has no vldb servers. Short circuit the error by testing for
1972 freelance in CreateFile and MakeDir
1974 * Treat srvsvc, wkssvc and ipc$ as pseudo files. Permit them
1975 to be opened and closed without actually existing. For now
1976 we treat them as equivalent to ioctls but this should
1979 * Fix RAP processing to not reject names that exist in the
1980 root.afs volume for GetInfo queries.
1982 * In afscreds, use GetUserNameEx to obtain the username instead
1983 of searching the registry. Try UserPrincipalName first and
1984 fallback to SamCompatible.
1986 * Stop linking to rpcns4.dll which is no longer supported on
1987 Vista and 2008. TaAfsAdmSvr is going to have to be
1990 * When processing dfs referral requests, if the sharename is
1991 a partial match and therefore will not be added automatically
1992 to the share list, make sure that we return an error.
1994 * Avoid a crash when logging VNOVOL errors returned from
1995 cm_CheckOfflineVolume.
1997 * When trace logging is disabled, do not permit osi_LogSaveString
1998 to perform the string copying. By disabling the string
1999 copies, throughput of the service more than doubles.
2001 * when checking offline volumes, update the cell vlserver info
2002 if required and don't perform a check if there are no servers
2005 * Add volumes in the vl_unknown state to the list of objects to be
2006 checked as offline or retried when an allbusy error is received
2009 * Add support for configuring rx_ExtraPackets
2011 * All calls to buf_Get() must be protected by a read lock on the
2012 cm_scache_t bufCreateLock
2014 * When a MergeStatus results in the invalidation of buffers, do
2015 not remove buffers from the hash tables that have active references.
2017 * Upon receiving a VNOVNODE error from a file server, be sure to
2018 invalidate any file locks. This prevents on-going attempts to
2019 extend the locks and will cause subsequent attempts to access
2020 the file to receive a invalid handle error.
2022 * Prevent smb_ReceiveTran2SearchDir from adding terminating NULs
2023 to directory entry names.
2025 * No longer discard tokens when RXKADBADTICKET error is received.
2026 Instead return STATUS_NO_KERB_KEY error.
2029 * Add "cmdebug <host> -cellservdb" option and client service
2030 implementation permits querying the vldb server for each
2031 cell known to the cache manager. The output is in a form
2032 that permits it to be used as a CellServDB file.
2034 * Add Symlinks->Show option to the AFS Explorer Shell Context
2037 * Prevent buffer overrun in the cmd parser used by command
2038 line tools such as fs, vos, pts, etc.
2040 * The path ioctl operations have several issues:
2042 (1) the specified path for "fs examine, whereis, whichcell,
2043 flush" and so always has the follow symlinks and mount
2044 points semantics. This makes it impossible to determine
2045 what the FID of a symlink or mount point is.
2047 (2) "fs examine" out is not the result of a single pioctl
2048 operation but is actually the combined output of half
2049 a dozen operations. Path evaluation is an expensive
2050 operation. It would be faster if the caller could
2051 evaluate the FID first and then perform all of the
2052 rest of the operations by specifying the FID instead
2055 (3) fs output reports all objects as files. By adding a
2056 GetFileType pioctl more informative output can be provided
2057 that indicates what type of object the path evaluates to.
2059 (4) the Windows fs command includes a number of commands that
2060 do nothing but exist only because the Unix cache manager
2063 A new extendible data structure cm_ioctl_query_opts_t has been
2064 added which can be optionally specified with pioctls that do not
2065 already require input data. The first two fields of this structure
2066 are 'literal' and 'fid'. The literal field is used to indicate
2067 whether the last component of the path should be evaluated following
2068 symlinks and mount points.
2070 The fid field permits a fid to be specified.
2072 * A new GetFileType pioctl has been added. The type of objects are
2075 * A new "-literal" option is available for "fs examine, flush, whereis,
2078 * Unimplemented fs commands have been removed.
2080 * There are circumstances where a volume object is being accessed and
2081 the volume is marked indicating that the volume location information
2082 is out of date but where it is also pointless and perhaps dangerous
2083 to block waiting for the rpc to complete. One example is when
2084 processing the cmdebug requests. If we know that we are not going
2085 to use the volume object to contact a server, then we can now set
2086 the CM_GETVOL_FLAG_NO_RESET flag.
2088 * when moving up in the directory tree we search the recorded fid list
2089 to find a matching fid that we have already crossed. we must also
2090 reset the fid count based upon what we discover.
2092 * Cell names are published as share names. As a result they are searched
2093 for as part of the Dfs Referral evaluation. If share "foo" can not be
2094 reached, the CIFS client will fallback to searching for "fo" as well.
2096 Since the freelance client automatically adds a symlink for prefixes
2097 of the cell name, this results in both "foo" and "fo" being added to
2098 the freelance root.afs volume. This patch prevents that negative side
2101 * when tracking Kerberos credential cache names be sure to include
2102 the cache type prefix.
2104 * Microsoft has assigned OpenAFS a network type value
2106 * The algorithms that were used to produce a Netbios name were broken.
2107 The name that was produced when the hostname was too long was in fact
2108 longer than the maximum permitted netbios name.
2110 Also, the "NetbiosName" value was not used as a suffix when the loopback
2111 adapter was not installed. The hardcoded string "AFS" was used instead.
2113 * avoid another deadlock during server probes initiated by the ipaddr
2114 change daemon thread.
2116 * prevent the afs client service from crashing when fs newcell is
2119 * In the NSIS installer, move afslogon.dll, afscpcc.exe and afs_cpa.cpl
2120 to \Program Files\OpenAFS\Client\Program in order to get them out
2121 of the %windir% tree
2123 * EDQUOT == WSAEDQUOT. Define it in the right places so that in the
2124 end STATUS_OUT_OF_QUOTA can be returned to the application.
2126 * Modify the search order for determining the location of CellServDB
2127 and other client configuration files.
2129 1. AFSCONF environment variable
2131 3. NEW - use All Users\AppData\OpenAFS\Client only if CellServDB exists
2132 4. use Program Files\OpenAFS\Client
2134 * More Vista Power Management tweaks
2136 * Close a socket handle leak in the kauth routines
2138 * Don't hold a cm_volume_t mutex across RPCs
2140 * Convert downTimes to use clock time instead of relative times.
2141 This permits correct comparisons with expiration times.
2143 * Update the Wix installer to properly install the prerequisites
2144 for translate_et.exe
2146 * Do not translate pioctl paths as they are not converted by the
2149 * Implement multi_rx probes for checking whether servers are up
2150 or down. Turns probes into a constant time activity instead
2151 of one dependent upon the number of known servers.
2153 * Treat VNOVOL errors the same as VOFFLINE. Do not force an
2154 update of the volume location data and retry. VNOVOL means
2155 that the volume could not be attached. It doesn't mean that
2156 the volume is not on this server.
2158 * Modify the Explorer Shell AFS Context menu Symlink->Add to
2159 not attempt to validate the target path. There is no reliable
2162 * Improve performance of \\afs\<cell><mount-type><volume>\
2163 access by eliminating extraneous attempts to search the
2164 CellServDB file and DNS AFSDB records. Share names
2165 containing '%' or '#' are not valid cell names.
2167 * Return STATUS_RANGE_NOT_LOCKED if the application attempts
2168 to unlock a range that is not currently locked.
2170 * Enable the 32-bit tools to read the 64-bit registry keys
2171 when executing in the WOW64 environment.
2175 * Fix buffer refcount leak introduced in 1.5.29
2177 * Prevent the NIM AFS Provider from crashing when configured to use
2178 Kerberos v4 for token acquisition in situations where no Kerberos v4
2179 support is available. For example, 64-bit KFW or sites that delete
2180 the Kerberos v4 DLL.
2182 * In afsd_init.log, support logging cache sizes greater than 2GB.
2184 * The volume status output of "fs examine" was being determined based
2185 upon the wrong error value. Instead of using the pioctl() return
2186 code, the errno value must be used.
2188 * Prevent invalid pts auto-registration attempts by aklog and the
2189 NIM afs provider when Kerberos referrals are in use. (krb5 1.6+)
2191 * Add synchronization protection to all of the SMB Listener State
2192 variables. This prevents race conditions when the addition or
2193 loss of a network adapter takes place.
2195 * On Vista, there is a race condition between the restart of the
2196 afs client service and the network adapters. Ensure that the
2197 afs client service can handle switching between loopback and
2198 non-loopback modes of operation.
2200 * Fix FollowBackupPath to work with mount points that are not
2201 explicitly read/write.
2203 * Fix a deadlock that has been observed on Vista while resuming
2204 after at least four hours of sleep.
2206 * Add new HKLM registry configuration options that can be used
2207 to provide configuration hints to the NIM AFS Provider when
2208 creating new identities. See release notes for details on
2209 HKLM\SOFTWARE\OpenAFS\Client\Realms\.
2212 * Speed up write buffer operations by holding a global read-lock
2213 instead of a write-lock when updating the buffer data version
2214 number. The data version is protect by the associated stat
2217 * Speed up read buffer operations by removing out of date buffers
2218 from the buffer hash tables.
2220 * Protect against a null pointer dereference in afslogon.dll
2221 if QueryAdHomePathFromSid() fails.
2223 * Remove a race condition while accessing private freelance data
2226 * Update the buffer data versions when Freelance directory data
2229 * Complete the implementation of smb_ReceiveTran2GetDFSReferral().
2232 * Return EAS_NOT_SUPPORTED in response to EA read/store/query
2233 instead of returning zero length.
2235 * Return ERROR_RETRY when a server sends VBUSY.
2237 * Return PATH_NOT_COVERED instead of BADSHARENAME for DfsLinks when
2238 a volstat plug-in is in use.
2240 * Enhance the volstat interface to support Dfs path mappings.
2242 * Fix "symlink list" to do the right thing for Dfs links
2244 * Fix callback expiration that was broken in 1.5.27.
2246 * Fix a crash at shutdown caused by the lack of synchronization
2247 within the daemon threads (cm_Daemon, cm_BkgDaemon, cm_IPAddrDaemon)
2249 * Hold the cm_server_t mutex while walking or modifying the
2250 cm_server_vols_t list in order to avoid a race condition.
2252 * Ensure that the cm_server_t refCount is safely modified.
2254 * The pthread library implementations of pthread_cond_wait
2255 and pthread_cond_timedwait were failing to return with
2256 the mutex held when error conditions occurred. This could
2257 result in race conditions wherever they are used. One case
2258 in which problems were seen was the Rx event_handler().
2260 * Prevent cm_GetBuffer() from locking all buffers associated with
2261 a stat cache object before deciding if the current buffer is up
2264 * Fix cm_BkgPrefetch to actually ensure that there is a cm_buf_t
2265 object for each offset that we will attempt to reserve.
2267 * Reduce the default number of background daemon threads used to
2268 prefetch or store to one in order to prevent two or more threads
2269 from stepping on each other.
2271 * Reduce the default chunk size to 256KB in order to prevent read
2272 timeouts on low bandwidth connections.
2274 * Fix a race condition that could result in threads becoming stranded
2275 when synchronizing multiple requests on the same stat cache object.
2278 * The GiveUpAllCallBacks functionality that was added in 1.5.21
2279 has been disabled. It has been disabled because it turns out
2280 that the RPC is improperly implemented in all file servers
2281 previous to 1.4.6. The RPC is executed without the proper
2282 locks being held resulting in data corruption and eventual file
2283 server failure. The functionality will be re-enabled in a
2286 * a couple of .readonly Volume Callback optimizations were
2288 - Apply most recent volume callback to all stat cache objects
2289 in the volume. This avoids premature callback expirations
2290 and unnecessary FetchStatus calls.
2291 - Add option to permit attempts to renew .readonly Volume
2292 Callbacks on a periodic basis to avoid invalidation of stat
2293 cache data. [Requires daemonCheckVolCBInterval be set to
2294 non-zero number of seconds. 1800 (30 minutes) is suggested.]
2296 * Adds support for 64-bit data version values. Previous releases
2297 only supported 32-bit values. Larger values sent by the file server
2298 would result in rollover.
2300 * aklog and asetkey no longer generate an exception if Kerberos for
2301 Windows is not installed.
2303 * If the LSA authentication functions report a temporary out of memory
2304 condition, fail the current authentication and do not cause the AFS
2305 client service to terminate unexpectedly.
2307 * When the AFS client service is shutting down, prevent the daemon check
2308 thread from executing additional checks which might attempt to access
2309 discarded locks or unmapped memory.
2311 * Correct a double free condition in the NIM AFS credentials provider.
2313 * Add "FollowBackupPath" service parameter registry value.
2314 This provides equivalent functionality to the UNIX afsd -backuptree
2315 option. When set to a non-zero value, normal mount points originating
2316 in a .backup volume will prefer a .backup volume.
2319 * Fix NSIS uninstall from the .MSI installer.
2321 * Add Stack Frame logging for AMD64
2323 * Add support for EWOULDBLOCK to cm_Analyze. If the file
2324 server returns EWOULDBLOCK, retry the request every two
2325 seconds for up to the RDRtimeout.
2327 * Reorganize the order of the includes to ensure that EWOULDBLOCK
2328 is not assigned the same value as EIO.
2330 * Update CellServDB to GCO Public 25 Oct 2007
2332 * Add additional validation and error handling code after
2333 each call to getSlot(). If an invalid slot is returned,
2334 return NONODE. If the invalid slot is returned when
2335 extracting a data node, invalidate the tree.
2337 * Modify compareKeys() to always perform a case-insensitive
2338 comparison and only perform a case sensistive comparison
2339 if the case-insensitive one matches. This ensures the
2340 ordering is consistently reported.
2342 * Add lock assertions to ensure that all calls are being
2343 performed with the correct locks being held. There have been
2344 some crash reports that provide stack data that does not appear
2345 to be possible unless there is a race. However, there are no
2346 obvious locations where the race is taking place and the test
2347 suite indicates that all of the correct locks are being held.
2348 We shall see what happens in the field.
2350 * For consistency replace all calls to findKey in which the range i
2351 s (1,numentries) with calls to getSlot().
2353 * Optimize the depth search loop by testing the slot value in the
2354 for statement instead of forcing the loop to be broken later.
2356 * Reorganize the locking for cm_BeginDirOp and cm_EndDirOp.
2357 There are a number of locations where locks are obtained, dropped,
2358 and reobtained. This reorganization attempts to accomplish several
2361 (1) be optimistic for the most common case so it will be fast
2363 (2) add consistency checks after each location where locks are
2364 dropped and re-obtained. If we lose a race in cm_BeginDirOp
2365 and the bplus tree is out of date, retry until we get to a
2366 consistent state that we can use.
2368 (3) Ensure that all operations take place with the correct locks.
2370 * One of the issues that has become a serious problem since the
2371 addition of the local directory updates is that although cm_SyncOp
2372 synchronizes operations, it does not preserve the order of requests.
2373 This has always been a problem in that it has been possible for a
2374 request to fail to complete due to its worker thread's bad luck.
2375 When a request takes longer than the Windows SMB Redirector's timeout,
2376 the SMB Redirector tears down the SMB virtual circuit.
2378 When using the local directory updates it is really important that
2379 the directory update operations complete in the order that they were
2380 sent to the file server. If they don't, then the local directory state
2381 and the file server state will not match and the local directory state
2382 must be discarded which in turn forces a new read of the entire
2383 directory contents over the network.
2385 This patch adds a new cm_scache_waiter_t object that is used to store
2386 the current thread, buffer, and syncop flags within a waiters queue
2387 on each cm_scache_t object. If a thread is forced to sleep in
2388 cm_SyncOp, upon waking it will check to see if there are any other
2389 threads waiting that are attempting to perform a similar task ahead
2390 of it in the queue. If yes, the thread goes back to sleep. If not,
2391 it goes ahead and enters the cm_SyncOp conflict resolution block.
2393 This patch has the additional side effect of reducing the number of
2394 competing threads that must obtain the cm_scache_t mutex and process
2395 the cm_SyncOp conflict resolution block. As a result, the overall CPU
2396 utilization of the service and the clock time associated with processing
2397 requests will be reduced.
2399 * assert that the cm_scache_t mutex is held when calling
2400 buf_ForceDataVersion().
2402 * rename findNode to leafNode in bplus_Lookup
2404 * replace all OutputDebugString calls with osi_LogX calls
2406 * modify bestMatch to special case the return values for leaf nodes.
2407 If an entry is above or below the values available in the leaf node
2408 return BTLOWER or BTUPPER instead of BTERROR.
2410 * In insert and delete operations check for BTLOWER/BTUPPER and isleaf,
2411 if true convert to either slot 0 or Max and perform the insertion.
2412 This produces easier to read code when performing lookups.
2414 * modify lock_AssertXXX macros to call osi_assertx() and provide a
2415 descriptive message.
2417 * the difference between osi_assert() and osi_assertx() is whether or not
2418 a description is specified as part of the call. When no message is
2419 specified afsd_notifier() is called with a NULL msgp parameter. This
2420 results in a null pointer dereference during the EventLog() call.
2422 As a result, none of the information describing the assertion is written
2425 This commit sets a default message for use when no other message is
2426 provided and it replaces all calls to osi_assert with osi_assertx and
2429 * Migrate B+ tree search key into thread local storage
2432 * Fix an uninitialized variable in lana_IsLoopback() which
2433 could result in a random determination that a loopback
2434 adapter is not a loopback adapter. This would result in
2435 a computed Netbios service name of %MACHINE%-AFS instead
2436 of AFS. This is a serious problem if the value computed
2437 by the AFS Client Service and the client process pioctl
2438 calls produce different results as it will result in a
2439 communication failure.
2441 * Fix a crash in the B+ Tree code that results when a file
2442 rename only results in a change of case.
2444 * Fix the pioctl path parsing. The previous algorithm
2445 broke the current directory of the client into two parts,
2446 the TIDpath which represented the device root and the
2447 relative path. The problem is that this is not sufficient
2448 from proper processing of symlinks and mount points.
2450 What is required is that the last component of the path
2451 must not have the symlink or mount point be evaluated
2452 whereas symlink and mount point evaluation must be performed
2453 for the rest of the relative path. Therefore, the path is
2454 now split into three parts: TIDpath, intermediate path,
2457 "fs examine" now reports the data of the symlink or mount
2458 point object instead of the target and "symlink list" and
2459 "symlink remove" now works. The "Explorer Shell Extension"
2460 menu now is displayed for symlink objects.
2462 * More informative symlink.exe error messages.
2464 * When the cmd.exe "del" command is used to remove a symlink,
2465 not only was the link object removed, but the stat cache
2466 object for the target was marked deleted. This resulted
2467 in cache corruption which would require the destruction
2468 of the cache to correct.
2470 * Add undocumented registry value, "OfflineReadOnlyIsValid"
2471 which results in cached contents from read only volumes
2472 to be treated as valid when there is no callback registered
2473 and all servers are down or all instances are offline.
2475 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
2477 * Add documented registry value, "AcceptDottedPrincipalNames"
2478 which is used to permit the acquisition of AFS tokens
2479 using Kerberos v5 principal names which include a dot in
2480 the first component.
2482 {HKLM\HKCU}\Software\OpenAFS\Client
2484 * "fs uuid" does not require "AFS Client Administrator" group
2485 membership to display the current UUID value.
2488 * Update NSIS installer to version 2.30
2490 * Update Wix installer to version 2.0.5325
2492 * 64-bit KFW support:
2493 - Network Identity Manager plug-in
2495 - 64-bit krb5 enable afscreds.exe, afs server manager,
2498 * Enable B+ tree Directory Search by default.
2500 * NIM plug-in is now compatible with FILE: ccaches
2502 * Install instloop.exe into Client/Programs/ even though
2503 it isn't required. Permit it to be used by organizations
2504 that clone Windows systems.
2506 * On a suspend, only drop callbacks for servers that
2507 are online. If the server is offline, maintain the callback
2508 until the server is once again available.
2510 * Prevent against acquisition of a directory write lock when
2511 there are no updates to be performed. This was resulting
2512 in a write-lock leak which later on would lead to a deadlock.
2514 * Add configurable file pre-fetch mode. For files that match
2515 the specified file extensions, pre-fetch the entire file into
2518 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2519 (MULTI_SZ) PrefetchExecutableExtensions
2521 * Fix the client_osi large integer support so that return values
2522 can be properly assigned to large integer variables.
2524 * Do not hold cm_scache_t mutex across cm_GetCell() calls
2525 since cm_GetCell() can block in a RPC.
2527 * Add short file name support to B+ tree searches.
2529 * Resolve a conflict between the desire to notify file servers
2530 that the client workstation is suspending and the desire to
2531 permit objects with valid callbacks when the client suspends
2532 to continue to use those objects from the cache if the client
2533 resumes without network connectivity.
2535 * Fix stack variables to ensure that all buffers storing paths
2536 can handle AFSPATHMAX (1024) characters. Paths of 256 characters
2537 would cause a stack overflow condition.
2539 * On Vista, force afs_config.exe to run with the highest privileges
2540 available to the end user. For a user in the Administrators
2541 group running under UAC, this will cause afs_config.exe to be
2542 be run with full Administrators privileges after the user
2543 approves the privilege escalation.
2545 * More VNOVNODE handling issues. If a dirty buffer cannot be
2546 written due to VNOVNODE, clear the buffer and ignore the error.
2547 If a cm_scache_t object has been marked deleted due to a VNOVNODE
2548 error, the next time the smb_fid_t object that references it is
2549 used return either NOSUCHFILE or NOSUCHPATH and close the
2550 smb_fid_t object. Subsequent attempts to use the smb_fid_t will
2551 result in an invalid handle error.
2553 * For Vista, set the RequestExecutionLevel of the NSIS installer
2556 * For the build environment, support the generation of manifests
2557 when using the CL1310 (VS .NET 2003) compiler/linker.
2559 * There was a bug in OutputDebugHexString() which could overflow
2560 a stack buffer resulting in unexpected process termination
2561 of afsd_service.exe if the input Security Token contained
2562 many bytes equivalent to the '%' char.
2564 * Add directory lookup statistics to "fs memdump" output.
2568 * Windows uses case-insensitive file name pattern matching
2569 but AFS is a case sensitive file system. The AFS3 directory
2570 format is block based, uses network byte order and
2571 includes a hash table for fast case sensitive lookups.
2572 This causes several problems for the Windows AFS client.
2573 (1) Traversing the directory blocks is cpu expensive
2574 (2) A hash table miss does not indicate that the desired
2575 entry does not exist.
2576 (3) Determining whether a non-ambiguous inexact match or
2577 the entry does not exist requires a linear traversal
2578 of the entire directory.
2579 These issues often result in 100% CPU utilization.
2581 These issues are addressed by building a modified B+ tree for
2582 each directory and then using the B+ tree for searches.
2584 This feature is disabled by default. To enable it set
2585 the "BPlusTrees" registry value to 0x01 in the AFS Client
2586 Service Parameters key.
2588 * Correct a null pointer dereference in the symlink recursion
2589 detection code. (Introduced in 1.5.22)
2591 * Add the Network Identity Manager AFS Provider to AMD64
2596 * A new registry value, "BlockSize", can now be used to adjust
2597 the size of the internal buffers used to store file data.
2598 The default is 4KB. When set, the values 1..1024 specify
2599 multiples of 4KB. Values > 1024 specify actual block sizes
2600 and must be multiples of 4KB. The actual block size in KB
2601 can not be larger than the chunk size which defaults to 1MB.
2603 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2606 * pts.exe and ptserver.exe built with -DSUPERGROUPS
2608 * Service Manager shutdown hints have been added. The hint
2609 specifies that it may take the AFS client service up to two
2610 minutes to shutdown.
2612 * Fix the recursion detection feature introduced in 1.5.22.
2613 The previous implementation broke the evaluation of paths
2614 using ../path or ./path. It also broke the evaluation of
2615 symlinks to absolute paths. For example, /afs/cell/path.
2617 * Fix an uninitialized variable in cm_MakeDir that could result
2618 in crash if the requested directory could not be created.
2620 * Fix a the VolumeStatusNotification log messages to save
2621 the input string in the circular log buffer so garbarge
2622 won't be output to afsd.log when the log is dumped.
2625 * Increase the SMB Maximum Path length to 260 characters
2626 from 256 as that is what CIFS supports.
2628 * On 64-bit Windows, use a unique description string for
2629 the 32-bit AFS Shell Extension.
2633 * In the AFS Admin Library, replaced all of the
2634 Ubik_Call(XXX_RPC,...) calls with Ubik_XXX_RPC(...). This
2635 allows parameter checking and proper type conversions to be
2636 performed. This is crucial when passing parameters that are
2637 64-bit wide on systems in which 'long' is a 32-bit value.
2638 Otherwise, the parameter list on the stack is interpretted
2639 incorrectly and the function not only doesn't succeed but it
2640 might corrupt memory and if lucky crash the application.
2642 The only program that we care about that uses this library
2643 is the AFS Server Manager. Now it doesn't crash.
2645 * Fixed smb directory searches for non-wildcard names. The
2646 optimized function should not have set the bulk in progress
2647 flag. This would adversely affect later directory searches
2648 that required a bulk search.
2650 * Detect infinite recursions (up to 512 stat cache objects
2651 in a chain.) If an infinite recursion is detected, return
2652 an error to stop the CIFS client from querying us forever.
2654 * Executables executed out of AFS name space can have their
2655 pages swapped out by Windows. Those pages will be reloaded
2656 when they are required. However, if the cache manager no
2657 longer has a callback and all of the file servers with
2658 the volume data are down, then the application will crash
2659 due to the inability to read the required page.
2661 We now track the down time of each server. If the callback
2662 expires after the server that issued it went down and all
2663 servers are down, we delay the expiration to permit the data
2664 in the cache to be continue to be used by Windows or an
2667 * The Windows AFS client has always had very poor performance
2668 with regards to directory management. In the case of
2669 directory updates caused by create, rename,and delete operations,
2670 the AFS cache manager will now perform local directory updates
2671 when possible in order to avoid re-reading the directory from
2672 the file server. This will be done in whenever both of the
2673 following conditions are true:
2675 - the data version on the directory increased by one as a
2676 result of the operation
2678 - all of the directory buffers necessary to make the change
2679 are present in the cache
2681 * The error message in the AFS Control Panel describing the
2682 need to be a Windows Administrator has been replaced with
2683 a message describing the need to be in the AFS Admins Group.
2685 * If the vldb server list is empty when attempting to evaluate
2686 a cell mountpoint it was possible for a crash to occur. This
2687 can only happen if there is an existing mount point and the
2688 is a failure attempting to load the server list from the
2689 CellServDB file or via AFSDB queries.
2691 * dirty buffers only have their dirty bytes stored to the file
2692 server instead of the entire chunksize
2694 * the default chunksize has been increased from 128KB to 1MB
2699 * changed the enum values for cm_serverRef_t state info to use a
2700 private name space to avoid collisions (srv_)
2702 * added a srv_deleted state for cm_serverRef_t objects. This
2703 state is set when cm_FreeServerList() is called with the
2704 CM_FREESERVERLIST_DELETE flag set. cm_FreeServerList() may
2705 not always delete the cm_serverRef_t from the list if it is
2706 still in use by another thread. the srv_deleted state means
2707 the object's contents are no longer valid and it must be
2708 skipped. It will be deleted the next time the object is
2709 freed and the refcount hits zero.
2711 * the srv_deleted state is also used when a file server reports
2712 either VNOVOL or VMOVED instead of marking the cm_serverRef_t
2713 as offline. This is done to prevent additional usage of the
2714 stale vldb data while waiting for the update volume request
2717 * added a state field to the cm_volume_t object (enum volstate
2718 vl_ name space) that maintains the state of the volume based
2719 upon the states of all of the cm_serverRef_t and cm_server_t
2722 * modified cm_UpdateVolume() to set the state of the cm_volume_t
2723 RW, RO, and BK to either vl_alldown or vl_online. There can't
2724 be any other states because cm_UpdateVolume() destroys any
2725 previous knowledge we might have had regarding busy or offline
2728 * modified cm_UpdateVolume() to update the volume name in the
2729 cm_volume_t to the volume base name if the previous value was
2732 * modified cm_FollowMountPoint() to check to see if the volume
2733 name is a volume ID and if so call cm_GetVolumeByID instead
2734 of cm_GetVolumeByName. This ensures that volume IDs are always
2735 looked up as numeric values. There is no longer a need to
2736 maintain a separate cm_volume_t containing the string representation
2739 * Added a flags parameter to cm_GetVolumeByName() and cm_GetVolumeByID().
2740 The first flag is a "CREATE" flag which is set by all existing
2741 calls. The flag is not set by calls to cm_GetVolumeByID() from
2742 the server probe code when volume status is being updated. We
2743 do not want the server probe operation to result in additional
2744 turnover in the cached volume data. The second flag is NO_LRU_UPDATE
2745 which is set when the server probe code updates the volume status.
2746 This flag will be used to prevent the server probe operation from
2747 changing the order of the least recently used queue.
2749 * Modified cm_GetVolumeByName to ensure that only one cm_volume_t is
2750 allocated for a given set of normal, readonly, and backup volumes
2751 regardless of whether or not the volume is accessed via name or
2752 ID number. The cm_volume_t namep field is always the base name
2755 * Added a new volume state, vl_unknown. This state is used as
2756 the initial state for all cm_volume_t when the cache manager starts,
2757 for each cm_volume_t at creation, and for each cm_volume_t when
2758 recycling. The cache manager does not know the state of all
2759 volumes in the world, only those that are in the cache and for
2760 which it has queried the VLDB and hosting file servers.
2762 * modified cm_GetVolumeByName() to initialize the state of a
2763 volume to vl_unknown. The actual state will be set when a
2764 cm_VolumeUpdate() call completes successfully.
2766 * changed name of scache hash table variables to avoid ambiguity
2767 when adding hash tables for volumes
2769 * fix a buffer overrun in sys\pioctl_nt.c pioctl().
2771 * modified cm_UpdateVolume() to handle the case in which there is
2772 no RW volume but there is are RO volumes for a given base name.
2773 This is done by querying for the ".readonly" volume name if the
2774 base name does not exist in the VLDB. We never query for the
2775 .backup name because under the current usage model a .backup
2776 volume may only exist on the server that the read-write volume
2777 is located. If there is no RW volume, there can be no .backup.
2779 * Added four hash tables for cm_volume_t objects to improve the
2780 search time of cm_GetVolumeByID(), cm_GetVolumeByName() and
2781 cm_ForceUpdateVolume(). One each for Name, RWID, ROID, and
2782 BKID. Three ID hash tables are necessary as long as it is
2783 desireable to maintain a single cm_volume_t containing all
2784 of the related RW, RO, and BK volume data. Having the RW and
2785 RO volume data in the same object is necessary for the
2786 implementation of cm_GetROVolumeID() which returns either the
2787 RO or RW ID depending upon the existence of RO volume instances.
2789 * Added a volume LRU queue so that volume reuse becomes fairer.
2790 This does not replace the all Volumes list which is used when
2791 it is desireable to walk a list of all the volumes whose order
2792 is not going to change out from underneath you which makes it
2793 safe to drop the cm_volumeLock.
2795 * handles volume hash table updates where volume name to
2796 volume ID number changes. The volume name remains
2797 constant in the cm_volume_t. if a vos rename is performed,
2798 the name of the volume will change and the volume IDs will be
2799 updated. Subsequent access to the old volume ID will create a
2800 new cm_volume_t with the new name.
2802 * Added a daemon thread operation to query the state of volumes
2803 listed as busy or offline. cm_CheckBusyVolumes() calls
2804 RXAFS_GetVolumeStatus() for each volume ID that is marked vl_busy
2805 or vl_offline. If the volume is now online, the status on the
2806 volume is updated. The default period is 600 seconds. This can
2807 be configured with the BusyVolumeCheckInterval registry value.
2809 * Added prototype for smb_IoctlPrepareRead() which was missing a
2810 return type in the function definition.
2812 * Added volume id lists to the cm_server_t. These lists are
2813 allocated in blocks of ~32 IDs. When a cm_PingServer()
2814 detects a change in server state, the state of the cm_volume_t
2817 * Added volID to the cm_serverRef_t object. volID is used
2818 to identify the volume for which the object is a referral.
2819 cm_FreeServerList() uses the volID to remove the volume
2820 from the cm_server_t.
2822 * In cm_Analyze, when VNOVOL or VMOVED are received,
2823 call cm_ForceVolumeUpdate() to force a refresh of the volume
2826 * Added cm_VolumeStatusNotification() which is used at the moment
2827 to log volume status changes to the trace log. It will also
2828 be used as the access point to the File System Filter driver
2829 notification engine.
2831 * Added an all cm_scache_t list to cm_data. This replaces the use
2832 of the stat cache LRU queue when we need to enumerate all
2833 entries. The LRU list order is not static and when using it to
2834 enumerate all entries it can result in items being missed or
2835 items being processed more than once.
2837 * Modified cm_Analyze(). Instead of reseting the busy or offline
2838 state of a volume and forcing a retry of the operation
2839 cm_Analyze will defer to the background daemon thread that will
2840 update the state once every 600 seconds.
2842 * Added the automatic generation of a Freelance ".root" read-write
2843 mountpoint that refers to the root.afs volume of the workstation
2844 cellname at the time the mountpoint is created.
2846 * re-write cm_Analyze to make better use of the known volume
2847 status. VL_Server queries cannot result in CM_ERROR_ALLOFFLINE
2850 * renamed cm_CheckBusyVolumes to cm_CheckOfflineVolumes.
2851 busy volumes will be reset to srv_non_busy by the function
2852 but there is no mechanism for querying the busy state other
2853 than by attempting to access the resource.
2855 * cm_Analyze will query the state of an offline volume before
2856 deciding whether or not to retry when all volume instances
2859 * the algorithm used after an IP address list change detection
2860 has been modified. we now check the state of the VLDB servers,
2861 then reset the rx connections, and finally check the state of
2862 the file servers. Previously we flushed the rx connections
2863 and then checked the servers in whatever random order they
2864 were first discovered.
2866 * fix the volume status reported by "fs examine"
2868 * various improvements to the AFS Network Identity Manager Provider.
2869 - Don't access freed objects after the window has been destroyed.
2870 - Write back the AFS cell list in the identity configuration
2871 even when its empty.
2872 - Disable all use of the provider when AFS is disabled for a
2873 particular identity.
2875 * Fix cm_cell_t leak. When evaluating a mountpoint that uses an
2876 alias instead of the full cellname, double check the full cell
2877 name and see if we have a matching cell before allocating a new
2880 * Dump more data with "fs memdump". All volumes and scaches and
2883 * Reverse the order of the all cells list. This ensures that
2884 the workstation cell is always first in the list.
2886 * If we recognize the server from which a callback revoke was
2887 received, then we can ensure that we only invalidate the status
2888 information for cm_scache_t objects belonging to that cell.
2890 * When building for amd64, use APPVER 5.02 as we do not support
2891 64-bit Windows 2000 and explicitly state the machine type in
2894 * no longer treat VMWare network adapters as equivalent to a
2895 loopback adapter. For some unknown reason, publishing the "AFS"
2896 netbios name on them frequently results in a name collision
2899 * Use read locks whenever possible instead of write locks when
2900 walking the allCellsp list. This permits better use of multiple
2903 * In 1.5.20 a bug was introduced by the fake status when PRSFS_READ
2904 is not present on the directory optimization. It would report
2905 mount points and directory objects for which there was no
2906 status information as a zero length file. Instead we can use
2907 the vnode number of the object to determine if it should be
2908 considered to be a directory or a file. If it is odd it is a
2909 directory; even it is a file.
2911 * Give Back Callbacks to the file servers in response to suspend
2912 and shutdown power management events. Only give up callbacks
2913 to UP servers and do so with a very short RX dead timeout.
2916 * Fix a race condition in the SMB FID cleanup code that could
2917 result in a stat cache mutex never being released.
2919 * Fix the graphics files that are installed as part of the
2920 IBM Administrators Guide.
2922 * Change the behavior of pioctl path parsing to not follow
2923 symlinks or mount points during evaluation. This results in
2924 the FID being the symlink or mount point and not the object
2925 they point to. This is consistent with the behavior of the
2928 * Enable AFSDB record lookup for command line tools such as
2931 * CellServDB update 2007-07-05
2933 * Added a Volume Status Plug-in Interface that permits third parties
2934 to receive notifications when the volume state changes occur.
2936 * Ensure that all writes to the afsd_alloc.log file use CR-LF.
2938 * Prevent the afsprocmgmt.dll from being unloaded while the
2939 background thread it starts is running.
2941 * Avoid a race condition when closing smb_fid_t objects.
2944 * fix install path of IBM Quick Start Guide when installed by
2947 * fix the client so that it can write into directories with
2948 only 'li' permissions.
2950 * fix installation of 32-bit Shell Extension for 64-bit systems.
2951 Install it under its own HKCR keys.
2953 * modify the wix installer to remove and upgrade install packages
2954 with the same version number.
2956 * when flushing volume location data every two hours or in
2957 response to "fs checkvolumes", also flush the mount point
2958 links in order to force their re-evaluation.
2960 * aklog was copying an uninitialized buffer as part of the
2961 foreign cell automatic pts registration
2963 * when obtaining directory listings, check for read permission
2964 on the directory. If read permission does not exist, fake the
2965 status information for the directory entries. This improves
2966 performance, prevents the client from pounding the file server,
2967 and prevents leakage of cached status information.
2969 * "fs getcacheparms" output was incorrect.
2971 * NetIdMgr AFS Provider should only automatically add token
2972 acquisition for cell foo.com if the realm of the cell matches
2973 the realm of the identity.
2975 * Fix a null pointer dereference in cm_MergeStatus()
2979 * network identity manager provider:
2981 - Correct for the fact that pr_CreateUser tampers with the
2982 contents of the username input buffer.
2984 - Take into account that the krb4 credential data structure
2985 does not have separate realm fields for the user and service.
2986 When krb524 is used with a cross-realm ticket, the realm
2987 describes the service not the user.
2989 * Many integrated logon changes:
2991 - Fix eventlog reporting. Do not attempt to log an event
2992 if the event source registration fails. Use DebugEvent0
2993 instead of DebugEvent when there are no parameters.
2995 - Modify the LOOKUPKEYCHAIN macro to recognize ERROR_MORE_DATA
2998 - Fix the reading of Domain specific configuration for
2999 LogonScript and TheseCells. Previously the dwSize value was
3000 being overwritten so that subsequent RegQueryValueEx call
3003 - Fix a memory leak in the TheseCells reading code.
3005 - Add support for Domain specific "Realm" specification. The
3006 realm is the realm to be appended to the username. When
3007 logging in as a domain or to the local machine, the specified
3008 "Domain" name is not going to be a valid realm name.
3010 - Construct a proper principal name based upon the domain
3011 specified realm for use in obtaining tokens with KFW.
3013 - If the domain specified "TheseCells" list includes the
3014 default cell, do not obtain tokens twice.
3016 * An unprotected call to the Kerberos v5 function,
3017 krb5_free_context(), can result in an invalid memory access
3018 exception if the contents of the krb5 profile, krb5.ini, cannot
3019 be parsed correctly. The error is located within the afskfw
3020 library which is used by afscreds.exe, afssvrmgr.exe, and
3021 afslogon.dll. When the error is triggered within afslogon.dll
3022 a blue screen condition can occur when the winlogon.exe system
3023 process terminates unexpectedly.
3027 * Do not stop the afs client service with an assert
3028 if the cell name cannot be determined when responding
3029 to a "fs whereis" command.
3031 * Integrated Logon (afslogon.dll). No longer attempt
3032 to preserve Kerberos 5 tickets for the logon session
3033 as that job is now supported by KFW 3.1 and above.
3035 * The Network Identity Manager plug-in will attempt to
3036 remove expired service tickets from the cache.
3040 * CellServDB update - GCO Public CellServDB 09 Mar 2007
3042 * AFS Server Manager and AFS Account Manager data cache
3043 moved from HKCU\Software\TransarcCorporation to
3044 HKCU\Software\OpenAFS to prevent crashes due to data
3047 * Reduce the default UP server probe time from 10 minutes
3048 to 5 minutes to match UNIX. This will not solve the NAT
3049 port mapping timeout issue because most NATs default to
3050 a value less that 5 minutes and this value ensures that
3051 the afs client server will start probes 5 minutes after
3052 the last probe set ended. The probe period for any one
3053 server is therefore longer than 5 minutes. The more
3054 servers in the list the longer the time between probes.
3056 * Before marking a server as down in cm_Analyze, retry the
3057 request using a new rx connection to the server. If the
3058 cache manager is behind a NAT, it is possible that the
3059 response from the server is being sent to the original
3060 address:port from which the rx connection was established.
3061 This will appear to the cache manager as the server being
3062 down when in fact it is not. Forcing the use of a new rx
3063 connection will result in the server responding to the
3064 most recent address:port assigned to the client by the NAT.
3066 If the server is in fact down, it will simply take a little
3067 longer to notice it.
3071 * Removing a directory that contained entries was failing with
3072 the wrong error code being returned to the CIFS client.
3076 * rxdebug exponential back-off. don't take down a server that
3077 is already struggling to respond.
3079 * Network Identity Manager plug-in registers help on the Help
3080 menu if it is supported by NIM. (KFW 3.2 and above)
3082 * Plug-n-play networking did not always work after coming back
3083 from a Windows Suspend. The AFS client service failed to
3084 re-compute the LAN adapter to bind to and instead attempted
3085 to bind to all LAN adapters. This would polute the public
3086 network with the name "AFS" which could lead to conflicts
3089 * Build installers using Microsoft Hotfixes provided at
3090 http://tinyurl.com/243svk for Visual Studio C RunTime Library
3091 Daylight Savings Time issues.
3093 * "fs checkserver" does not require admin privileges unless
3094 the probe interval is being changed.
3098 * A reference leak was discovered in the mountpoint traversal
3099 functionality which resulted in cm_volume_t objects not becoming
3100 eligible for re-use. This will trigger a panic condition when
3101 more than the maximum number of volumes is accessed.
3103 * Fixed the success test for the "fs memdump" operation. Success
3104 is now reported as such. Added cm_volume_t object information
3107 * A data corruption error has been fixed. Write requests
3108 queued for background processing were not retried upon failure.
3109 This leads to inconsistencies between the contents of the file
3110 in the local cache and the contents of the file on the file
3113 The implemented fix is to re-queue any request that has failed
3114 due to timeout, volume busy, volume offline, server busy, or
3115 server offline. Requests are not processed while all servers
3116 for a volume are inaccessible. When servers become accessible
3117 the background request will be retried.
3119 In addition, if a background write fails, do not mark the buffers
3120 as clean. Failure to preserve the dirty flag prevents the buffers
3121 from being written to the file server in the future.
3123 This fix does not address the possibility that the credentials
3124 necessary to perform a write request might expire between the
3125 time the request is queued and when the request can be processed
3126 due to server accessibility.
3128 * The Freelance volume does not support per user ACLs. The
3129 cm_HaveAccessRights function did not handle this special case.
3130 Requests for privileges greater than read|lookup would result
3131 in afsd_service.exe entering an infinite loop. ISOBuster is
3132 known to cause this behavior.
3134 * When deleting files, the SMB Server would attempt to obtain
3135 the access permissions for the parent directory instead of
3136 for the file itself. This would result in an incorrect
3137 assessment of whether or not the file can be deleted.
3139 * The Network Identity Manager AFS plug-in has had its string
3140 tables updated for consistency with the language used by NIM.
3141 All credentials are referred to as credentials instead of tokens.
3142 Drop-down combo boxes now in fact drop down.
3144 * The AFS Salvager was not properly being built on Windows.
3145 This is relevant only if the AFS servers are in use.
3147 * The default signal handlers for the AFS servers incorrectly
3148 processed SIGQUIT. Instead of stopping the server, an
3149 exception was raised.
3151 * Fix another volume refcount leak. This one was triggered by
3152 an inability to contact the vlservers in order to update the
3155 * Fix checklist control used by the drive mappings tabs in
3156 afscreds.exe and afs_control.exe.
3158 * Improve output of "fs memdump" related to buffer management.
3160 * Fix refcount leak of cm_buf_t objects when prefetching buffers
3161 already in the process of being fetched.
3163 * Improve cache performance by computing the buffer hash table
3164 size based upon the number of buffers so that the average
3165 number of entries in a bucket is seven.
3167 * Improve cache performance of read/write data by locally
3168 updating the dataVersion of all buffers associated with the
3169 file on each StoreData or StoreStatus operation provided that
3170 the dataVersion was incremented by one and the dataVersion of
3171 the buffer matched the original dataVersion of the cached
3172 status information for the file.
3174 * When recycling cm_scache_t objects, make sure that we
3175 clear all of the previous fields. It is no longer the
3176 case that we can assume that the stat cache object is being
3177 obtained only after a successful FetchStatus call.
3178 InlineBulkStatus will allocate stat cache objects for which
3179 the no status information is available.
3181 * When a file is being opened with the OPEN_ALWAYS disposition,
3182 do not require write permissions when the file already exists
3183 unless of course the file is being opened requesting write
3186 * When a file has been opened for read-only access, if the
3187 application requests an exclusive lock, obtain a read-lock
3188 instead of a write-lock. The purpose of the exclusive lock
3189 in this situation is to prevent changes to the file and the
3190 full file read lock will provide this guarrantee.
3195 * Enforce Unix Mode Write bit to prevent file deletion of read only
3198 * The CIFS READ and LIST permissions use the same value 0x0001.
3199 When a directory is being accessed, check for PRSFS_LOOKUP
3200 instead of PRSFS_READ.
3204 * Fix computations of short names of mount points. They were
3205 broken as part of the directory search optimization. (1.5.13a)
3207 * When all buffers are in use, periodically sleep the threads
3208 waiting for a buffer to become free while the global buffer
3209 lock is not being held in order to permit those threads that
3210 can free buffers to do so.
3212 * Power Management improvements. Maintain a global flag that
3213 specifies whether or not the service is in a suspend state.
3214 Do not panic if Netbios() returns NRC_BRIDGE meaning that the
3215 lana is no longer valid. Instead, stop the listener threads
3216 and if all listener threads are stopped, reset the lana_list.
3217 Allow the cm_Daemon() thread to periodically check the state
3218 of the smb listeners. If they are all stopped and the service
3219 is not suspended, attempt to restart them. If there are no valid
3220 lanas, return to the stopped state.
3222 * CreateX and NTCreateX use cm_CheckNTOpen() to test whether or not
3223 the user is permitted to obtain read or write locks. This function
3224 would obtain the lock and then drop it returning whether or not
3225 the lock could be obtained. If the lock was in fact required,
3226 CreateX/NTCreateX would then obtain it with cm_Lock(). The problem
3227 of course being that this pattern results in three RPCs to the
3228 file server (lock, unlock, lock). This is reduced to one RPC by
3229 implementing cm_CheckNTOpenDone() which frees the allocated byte
3230 range lock from cm_CheckNTOpen() after the cm_Lock() call is
3233 * Optimize the SMB FindFirst processing for the case in which there
3234 are no wildcards. If a cm_Lookup succeeds, generate a response
3235 from the stat cache entry we have a callback for. Its cheaper
3236 than walking each buffer of directory data associated with the
3237 directory stat cache entry when we have thousands of directory
3240 * Increase the SMB Ioctl MaxData size to support larger token sizes
3242 * Disable autorun of the AFS Server Wizard when the AFS Server is
3245 * Add AFS Server service dependencies on tcpip and PNP_TDI
3247 * Remove the AFS Server Configuration Tool shortcut
3249 * Add Freelance mode awareness to AFS Server Configuration Tool
3253 * Unlike previous versions of the OS, Vista performs a shutdown on
3254 the Microsoft Loopback adapter just like it would on a real adapter.
3255 This causes the smb_Listener threads to trigger a panic during a
3256 suspend/hibernate power event.
3258 The fix is to unbind from the network adapters in response to a
3259 suspend/hibernate power event and then rebind to the adapters
3260 when a resume power event is received. Note that the resume events
3261 are not reliably delivered so it is possible the afs service will
3262 not be accessible. However, this is the best we can do.
3264 * don't allow environment variables we set in afslogon.dll to be
3265 inherited by child processes
3267 * force an rx connection reset if VICECONNBAD or VICETOKENDEAD are
3270 * The old dirty buffer synchronization algorithm had a buf_IncrSyncer
3271 thread walking the all buffer list periodically searching for dirty
3272 buffers to write to the file server. This had several negative
3273 results. The algorithm ate up ever increasing amounts of CPU time
3274 even when AFS is idle as the size of the cache increases. Also,
3275 buffers were written to the file server in an order based upon the
3276 original buffer allocation which has nothing to do with the order
3277 in which the buffers became dirty.
3279 The new algorithm maintains a dirty buffer list. Items are added
3280 when the buffer is originally marked dirty. A buffer is only
3281 removed from the list by the buf_IncrSyncer when the buffer is no
3282 longer dirty. If the list is empty the thread goes back to sleep
3283 immediately without additional processing requirements.
3287 * Enable renaming of directories from the Explorer Shell when
3288 Offline Folders is enabled
3290 * An error is reported in response to the SMB ReceiveCoreFlush
3291 request when an error has not occurred.
3295 * optimize dirty buffer flushes in response to suspend operations
3296 do not flush if only loopback adapter is installed
3297 do no retry dirty buffer writes
3299 * do not reset netbios when performing loopback checks. doing
3300 so will break the existing lana bind
3302 * remove a deadlock in smb3.c QPathInfo
3306 * Add NetIDMgr AFS Plug-in
3308 * "Show Tray Icon" checkbox removed from afs_config.exe
3310 * Disable DNS Registrations for the Loopback Adapter and make sure
3311 that Netbios is turned on.
3313 * A reference count leak on stat cache objects was fixed
3315 * For debugging, tread ids of writers are now tracked on locks
3317 * Corrected an incorrect lock that was being held but not
3318 freed in a rare error case
3320 * Changed trace log end of line to CR-LF so it can be read
3323 * modify the buffer synchronization thread to write all dirty
3324 buffers and sleep only when there is no work to be done
3326 * improve behavior of cache manager when large numbers of
3327 temporary files are being created by multiple clients within
3330 * do not permit dirty buffers that belong to deleted files
3331 to be written to the file server. its a waste of bandwidth
3333 * do not leak stat cache references when the file server
3336 * afscreds: do not permit an attempt to obtain tokens if the
3337 afs service could not be started. this was causing stack
3338 overflows on Vista after resuming from a suspend
3342 * better handling of dot directory
3344 * fs sq and fs mkm now behave as on UNIX
3346 * add license to installers
3348 * Return Invalid Parameter error for unknown InfoLevels in
3349 SMB Tran2 Search Directory.
3351 * Correctly use WSAStartup and WSACleanup within the RX library
3353 * Check that afsd_service.exe is running before performing
3356 * Force AFS Shell Extension to load resource DLL using
3357 afs_shl_ext.dll as a base name.
3359 * (Wix) install afs_shl_ext_XXXX.dll and afs_creds_XXXX.dll in
3360 OpenAFS\Client\Program
3362 * Fix SMB_SET_FILE_END_OF_FILE_LENGTH
3364 * Do not panic if the maximum number of volume entries are in use
3365 and one of them can be recycled.
3369 * Fix the code that reads the backconnnectionhostnames value
3370 so that there is not an uninitialized variable
3374 * integrated logon could cause winlogin.exe to blue screen windows
3375 if it was called with a domain that does not exist on the machine.
3376 this is possible during a sysprep.
3378 * corrections to cm_buf_t handling to ensure that buffers are
3379 removed from queues before they are freed and that holds are
3380 obtained on the next buffer in a chain before releasing the
3383 * do not generate MessageBox dialogs on network errors. they
3384 cause more trouble then they are worth.
3386 * move GlobalAutoMapper into a separate thread so that it does
3387 not prevent service startup if the WNetAddConnection2 call
3390 * restrict hard dead and connection timeouts to the CIFS Session
3391 Timeout value. This is being done to try to prevent virtual
3392 circuits from being destroyed prematurely.
3394 * Fix the code that writes the backconnectionhostnames value
3395 to ensure that the data buffer is written with the correct
3398 (there were no windows specific changes between 1.5.4 and 1.5.5)
3402 * Corrections to the Integrated Logon functionality that permits
3403 Kerberos 5 TGTs to be communicated into the user session. The
3404 transfer is now more secure and succeeds when the user does not
3405 have read permission to the %SystemRoot%\TEMP directory.
3407 * NSIS installer updated to version 2.18
3409 * WiX installer updated to version 2.0.4221.0
3411 * Updates to the IFS functionality for 64-bit correctness and
3412 permit compilation. The IFS functionality works for pioctl()
3413 operations but not much else at this point.
3415 * A major re-write of the OpenAFS implementation of the CIFS Tran2
3416 query and set functions.
3418 * Ignore the Windows CIFS client session timeout entirely. Timeout
3419 only after the AFS Hard Dead Timeout period expires.
3423 * move headers, libraries, and samples from OpenAFS\Client to
3426 * remove the auto-registration of AFS IDs for foreign cells
3427 from integrated logon. this permits afslogon.dll to avoid
3428 initializing the RX library. afslogon.dll is a network
3429 provider library and as such it is loaded into all applications
3430 which perform a CreateProcessAsUser call. Initializing the
3431 RX library spawns background threads to listen for inbound
3432 packets. When the network provider is unloaded from the
3433 process, the RX library does not cleanup the threads. If they
3434 continue to execute after the library is unloaded, the process
3437 * implement inline bulk status calls. this will significantly
3438 improve the performance of directory listings.
3440 * prevent a crash when evaluating mountpoints to volumes that do not
3443 * new command: fs uuid [-generate]
3444 displays the current uuid and allows a new one to be generated
3445 on the fly without restarting the afs client service.
3447 * fix the 64-bit installers to properly document the install
3450 * move help files to the proper location so they can be located by
3451 the system tray tool and the control panel.
3453 * "over quota" and "disk full" errors were not being reported to the
3456 * largefile support has been added. Files greater than 31-bits can no
3457 longer be written to file servers which do not support large files.
3458 When file servers support large files, files up to 64-bits can be
3461 * cm_SyncOp could return with the input buffer unlocked upon error
3462 which would result in a panic condition. This bug was introduced
3465 * vlserver failover when freelance mode did not work. This prevented
3466 the root.afs volume from being loaded if the first vlserver could
3471 * A bug in the CIFS Standard Info response packet caused directories
3472 to appear as standard files. This prevented Microsoft Vista from
3473 being able to walk the AFS directory tree.
3475 * Mount points must be reported as directories in response to
3476 Query Path Info operations.
3478 * A deadlock was detected when performing "fs flushall" if the
3479 file server reports VNOVNODE. The scp->createBufferLock is
3480 already held by the current thread and there is no mechanism
3481 to propagate the knowledge. Therefore, an alternate mechanism
3482 for clearing the cache must be developed. A new function
3483 cm_RecycleSCache(scp,flags) has been extracted from
3484 cm_GetNewSCache(). This function performs the task of recycling
3485 an cm_scache_t object. When called from cm_GetNewSCache() with
3486 no flags the expectation is that there are no associated buffers
3487 that are queued to be read or written. When called from
3488 cm_Analyze() with the CM_SCACHE_RECYCLEFLAG_DESTROY_BUFFERS
3489 flag, any queued buffers will be de-queued and marked as if
3490 the operations were performed so the data can be discarded.
3492 * CM_ERROR codes were logged as "unknown" by cm_Analyze.
3494 * In response to VNOVNODE, the parent is only discarded if the current
3495 cm_scache_t is not a directory.
3497 * In the Ioctl FlushFile and FlushVolume functions, there were no
3498 checks to protect against flushing the Freelance SCache entries.
3500 * In FlushFile, the wrong cm_scache_t object was being released.
3502 * In cm_GetNewSCache, do not allow Freelance SCache entries to be
3503 recycled. Choose a new entry if cm_RecycleSCache fails.
3505 * Speed up the performance of the cache manager by not
3506 holding the smb_fid_t mutex across calls to cm_SyncOp
3509 * Ensure that all smb_fid_t flag references are protected.
3511 * Remove a deadlock between smb_fid_t mutex and
3512 smb_rctLock introduced in 1.4.1.
3514 * Fix aklog AFSID lookup to use not send the realm
3515 name when the realm is the local realm for the
3516 cell. The PT Server doesn't strip the realm
3517 name when it is the local realm.
3519 * Treat "\\afs\*." as an alias for "\\afs\all"
3521 * Dynamically adjust the priority of server threads
3522 based upon the age of the CIFS request that is
3523 being processed. The priority is bumped one level
3524 for every 15 seconds of processing time.
3526 * The original openafs contribution mis-used the
3527 cm_GetCallback function. This function should only
3528 be called as a side effect of a call to
3529 cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK). The way it
3530 was being called results in a FetchStatus RPC being
3531 performed even when the client already has a valid
3532 callback and does so in a manner that destroys the
3533 synchronization of the threads that are calling
3534 cm_SyncOp. If nothing else this results in a
3535 significant performance penalty.
3537 * More changes to cleanup of smb_vc_t objects in
3538 order to prevent race conditions.
3540 * Reduce the amount of Power Management event logging
3543 * If the client detects that the IP address has
3544 changed, force the use of new RX connections. This
3545 enables the file server to recognize that the client
3548 * Remove a race condition and deadlock associated
3549 with cleanup of smb_vc_t objects.
3551 * Fix the ANSI filename support. One function that
3552 required it was missing the translations.
3554 Changes not present in 1.4.1 and since 1.5.0: 1.5.1 released
3557 * Add a DllMain() function to afspthread.dll. Perform
3558 cleanup of thread local storage, rx queues, critical
3559 sections, event handles, etc. This was implemented
3560 to avoid crashes in afspthread.dll when applications
3561 which load/unload the dll attempt to terminate.
3563 * Lock request behavior has been altered. As of this
3564 release a locally allocated lock will be used whenever
3565 the volume is RO or the requesting user's permissions
3566 are 'rl'. This will allow end users to execute programs
3567 from RW volumes when they do not have the 'k' privilege.
3569 Note that a recent discovery indicates that the 'k' bit
3570 is only supposed to control the ability to obtain read
3571 locks. Write locks are to be controlled by the 'w' and
3574 * Code has been added to enable the cache manager to be
3575 able to obtain the AFSID of the authenticated user.
3576 This is required to be able to determine when "Insert"
3577 privileges should apply. AFS considers a file to be
3578 a new file whenever the owner's AFSID matches the
3579 AFSID of the user. The current implementation relies
3580 on the name specified during the SetToken call to be
3581 correct. This will be replaced in the future with a
3582 new RPC that allows the AFSID of the authenticated
3583 user to be obtained.
3585 * A 32-bit-tools installer for use on 64-bit systems
3586 is now available. These tools are for use with 32-bit
3587 versions of Kerberos for Windows or JPSoftware's 4NT.
3589 Since 1.4.1 rc10: 1.4.1 final released (9 April 2006)
3591 * Remove the use of the "active_vcp" variable. This variable
3592 no longer serves any useful purpose and was the cause of
3593 race conditions that produced under counts of the smb_vc_t
3596 * Change default minidump output to include data segments
3598 * restore access to cm_IoctlGetVolumeStatus and
3599 cm_IoctlSetVolumeStatus calls for AFS pioctl operations.
3600 Still need to figure out why they are being called
3601 for SMB/CIFS IOCTL operations instead of Trans2GetFileInfo.
3603 Changes not present in 1.4.1: 1.5.0 released (16 Feb 2006)
3605 * Update build system to include the "sys_name" in the
3608 * optimize the number of calls to obtain the smb_user_t for a
3609 smb virtual circuit.
3611 * add logging of server addresses, status and capabilities when
3614 * prevent multiple threads from pinging the same server simultaneously.
3615 instead, threads block and then use the result of the ping that was
3616 already taking place.
3618 * Add registry values to allow the defaults for daemon checks to be altered
3619 at startup. Log values to afsd_init.log.
3621 daemonCheckDownInterval
3622 daemonCheckUpInterval
3623 daemonCheckVolInterval
3624 daemonCheckCBInterval
3625 daemonCheckLockInterval
3626 daemonCheckTokenInterval
3628 * Add EventLog messages for transitions in fs crypt state.
3630 * Byte range locking (when locks will be obtained from the
3631 is still under discussion)
3633 * 64-bit Windows operating systems now supported
3634 (XP SP2 64-bit, 2003 SP1 64-bit, 2003 R2 64-bit)
3635 This is a native 64-bit implementation of OpenAFS for Windows.
3636 There is currently no 64-bit version of KFW and therefore
3637 there is no (easy) method for obtaining AFS credentials.
3638 32-bit KFW 3.0 can be used with 32-bit OAFW libraries
3639 to obtain AFS tokens but there is not for the feint of
3642 * A faster implementation of the fcrypt algorithms improves
3645 * The afsd_service.exe now supplies Event Log Message Text.
3646 The event log no longer generates warning text due to lack
3647 of message configuration.
3649 * When afs servers are unreachable and network stack returns
3650 a WSAEHOSTUNREACHABLE error, the AFS client will immediately
3651 mark the attempted server as being down. This allows for
3652 very fast response to the calling application. This change
3653 has a pronounced improvement when the Windows Explorer is
3654 used on disconnected mobile computers.
3656 * Added DebugOutputString debugging of the RX library
3658 Since 1.4.1 rc9: 1.4.1 rc10 released (10 March 2006)
3660 * Fix the rx-lwp implementation to randomly generate rx call
3661 identifiers which are supposed to be unique at a given point
3662 in time. This affects some of the command line tools used
3663 for debugging but not the AFS Client Service itself which
3664 uses the rx-pthreads implementation.
3666 * xstats_fs_test.exe and xstats_cm_test.exe are now built for
3667 Windows although they are not distributed
3669 * don't generate an exception when logging unexpected SMB packets
3671 * the data structures used by cm_IoctlGetVolumeStatus and
3672 cm_IoctlSetVolumeStatus are not compatible with the CIFS
3673 Get/Set File Information calls so don't bother executing them.
3677 * Fix data written to registry as part of BackConnectionHostnames
3680 * Fix initialization bug when allocating cm_server_t objects that
3681 would cause servers to appear to be down after each allocation
3682 or renewal. Server would appear up after the next check servers
3687 * Windows will return WSAECONNRESET to the next receive packet
3688 operation on a socket if an ICMP packet is received in response
3689 to a packet that was sent and could not be delivered because
3690 the port was unreachable. We have to treat this as
3691 WSAEWOULDBLOCK so that we can retry the select() check and
3692 not mark the connection as bad.
3694 Since 1.4.1 rc6: 1.4.1 rc7 released (14 Feb 2006)
3696 * The client manager maintains a least recently used list of
3697 all the ACLs. If an ACL is already the most recently used
3698 do not bother to remove it from the list and place it on the
3701 * As part of the process of adding a new server, set the
3702 server is down flag and then perform a ping. If the server
3703 is up, the flag will be reset. If the server is down, the
3704 client will not timeout unnecessarily while attempting to
3707 * If a Mountpoint String is the empty string, return PATH NOT
3710 * Modify behavior of the Network Provider to avoid querying the
3711 profile location if integrated logon is known to be disabled.
3713 * Correct behavior of 'LogoffPreserveTokens'. When set, tokens
3714 will always be preserved regardless of whether or not integrated
3715 logon was used to obtain tokens. When unset, tokens will be
3716 destroyed regardless of whether integrated logon was used to
3717 obtain the tokens unless the profile is stored in AFS.
3719 * Disable LogoffTransferToken functionality. The SMB V3 Logoff
3720 message is being delivered even when the user is not logging
3721 off of the computer.
3723 * Fix a deadlock condition introduced in rc6.
3725 * Cell names are case-insensitive. Since the founding of OpenAFS
3726 there has been a bug that treated "cell", "Cell", and "CELL" as
3727 different cells. This would result in what appeared to be
3728 random disappearance of user tokens or a failure to authenticate
3729 even when tokens were listed by the "tokens" command.
3731 Since 1.4.1 rc5: 1.4.1 rc6 released (2 Feb 2006)
3733 * Additional auditing of the smb code revealed that smb ioctl
3734 operations were failing to free smb_fid_t objects
3736 * The File Identifiers assigned to smb_fid_t objects could be
3737 given the value 65535 which is 0xFFFF which is also the
3738 INVALID_HANDLE value. Both 0 and 65535 are now treated as
3741 * Audited all assignments to fields and ensured that they are
3742 protected by obtaining the relevant mutex object.
3744 * Re-wrote the VNOVNODE handling to avoid a potential deadlock
3745 if both the current cm_scache_t object and its parent are
3746 both returning VNOVNODE.
3748 * Removed all the logic associated with the 'dead_vcp' variable
3749 which must have been added for debugging because its used
3750 was never thread safe and it truly served no purpose.
3752 Since 1.4.1 rc4: 1.4.1 rc5 released (25 Jan 2006)
3754 * Re-enabled LogoffTokenTransfer and LogoffTokenTransferTimeout
3755 registry settings. Default is ON and 120 seconds. If set to
3756 OFF, tokens will not be destroyed.
3758 * Fixed a number of race conditions and reference count problems
3759 associated with global queues.
3761 * Added a method to detect abandoned smb virtual circuits so their
3762 associated file handles, locks, user tokens, and smb session
3763 resources can be released.
3765 * Added a mechanism to allow user tokens to be preserved
3766 during integrated login and be freed after logoff.
3768 * Check access permissions to the file before the file is
3771 * More improvements to the handling of SMB Virtual Circuits
3772 when SMB sessions are closed.
3774 * Correct a race condition introduced in rc4 that could result
3775 in a crash of afsd_service.exe.
3777 Since 1.4.1 rc3: 1.4.1 rc4 released (11 Jan 2006)
3779 * Properly handle pre-mature SMB session closures.
3780 File handles, tree connections, and user sessions are now
3781 freed when the virtual circuit is closed.
3783 * smb_username_t and smb_user_t objects are now properly reference
3784 counted and freed as appropriate.
3786 * disable to use of AFS file locks. byte range locking is still
3787 implemented but all locks are managed locally. This prevents
3788 multiple processes from editing the same file on the same
3789 machine but it does prevent multiple processes on separate
3790 machines from modifying the file.
3792 Since 1.4.1 rc2: 1.4.1 RC3 released (4 Jan 2006)
3794 * A fix to "fs setserverprefs -vlserver". Multiple calls with
3795 the same server parameter would result in a panic condition
3796 if the server had not already been contacted by the AFS
3799 * Better warnings are written to the afsd.log file when
3800 the Windows RPC Protocols are not properly configured as
3801 is often the case on Windows 2000.
3803 * fix the procmgmt library to restore C RTL signal handlers when
3804 it is being detached from the process prior to process termination.
3806 Since 1.4.1 rc1: 1.4.1 RC2 released (30 Nov 2005)
3808 * close a lock leak in the rx library that could be triggered while
3811 * prevent removal of drive mappings in response to afscreds.exe -M
3813 * add logic to process VNOVNODE in cm_Analyze. Force re-evaluation of symlink
3814 strings and flush the stat cache entry.
3816 * prevent server objects from being freed if user preferences are set.
3818 * Update WIX support to 2.0.2419.0
3820 * Release all locks on the server at service shutdown.
3822 * A down server should be marked up as long as it responds to the client.
3823 It doesn't matter if the server returned an error as long as the error
3824 is something other than VRESTARTING. If there is such an error, all
3825 connections on that server should be marked for replacement.
3827 * Change the check "up" servers period from once per hour to once every
3828 ten minutes. This matches the behavior of the Unix cache manager.
3830 Post 1.4.0 revisions: 1.4.1 RC1 released (21 Oct 2005)
3831 * Byte Range Locking support has been added. Architectural documentation
3832 is located within src/WINNT/afsd/cm_vnodeops.c. Best efforts are
3833 made to enforce Windows mandatory locking model using locally managed
3834 byte range locks with full file locks obtained from the file server.
3835 Read-only volumes provide the user with an implied locking privilege.
3836 Read-write volumes require that a user be granted the locking privilege
3837 in order for a lock to be obtained. As most Windows applications use
3838 Shared file access modes, they require the ability to obtain locks in
3839 order to access files.
3841 * Implement error translation functions to convert Universal Error Codes
3842 output by modern servers to local errno and winsock error values.
3843 This fixes the error message problems using pioctl() calls from fs.exe
3844 and the AFS Shell Extension.
3846 Since 1.4.0 rc8: 1.4.0 final tagged (20 Oct 2005); released (31 Oct 2005)
3847 * Allow the AFS Client Service to set the firewall configuration by
3848 retrying until after Windows determines the boot period is over.
3850 * Do not attempt to configure the firewall on Windows 2000, XP pre-SP2,
3853 * Detect IP address changes from within the AFS Client Service for the
3854 purpose of checking down servers sooner
3856 * If the file server or vldb server report RXKADUNKNOWNKEY, return
3857 SEC_E_NO_KERB_KEY to Windows. This provides an indication to the
3858 user that there is a Kerberos key problem. Previously the user would
3859 receive no feedback when there was something wrong.
3861 * Fix the large integer library support to properly handle values greater
3864 Since 1.4.0 rc7: 1.4.0 rc8 released (15 Oct 2005)
3865 * Updated CellServDB
3867 * Fixed file name pattern matching. Question marks can match nothing
3868 if found at the end of a component.
3870 Since 1.4.0 rc6: 1.4.0 rc7 released (8 Oct 2005)
3871 * Kerberos 5 ticket lifetimes should adhere to KFW Leash lifetimes
3873 * The MSI installer now has a new Product ID for each build. This
3874 is to allow for automated upgrades. The Upgrade ID is remaining
3877 * Volumes that were cached and then subsequently were salvaged,
3878 moved or otherwise taken temporarily offline (as opposed to being
3879 busy) would become unreachable. A distinction is now made between
3880 a volume being offline and a server being down.
3882 Since 1.4.0 rc5: 1.4.0 rc6 released (2 Oct 2005)
3884 * Fixed several AFS Shell Extension issues:
3885 - Checking Server Status would only work if a specific cell was
3887 - If servers were down, the Server Down dialog would write to
3888 invalid memory and cause explorer.exe to crash
3889 - Default Push Buttons were not specified for many dialogs
3890 - The extension would be unloaded by the Explorer Shell and
3891 never be reloaded under some circumstances.
3893 * Fixed the AFS System Tray menu to disappear if another Window
3894 was selected before a menu item was selected
3896 * Added volume owner and group information to "fs examine" output
3898 * Added fs and registry support for enabling/disabling/clearing
3899 rx statistics gathering.
3901 Since 1.4.0 rc4: 1.4.0 rc5 released (25 Sep 2005)
3903 * New Product Code GUIDs issues for 1.4 and 1.5 branches. Separate
3904 GUIDs for each platform.
3906 * Documentation directory structure and default installation rules
3909 * New OpenAFS for Windows HTML Release Notes installed
3910 as part of the documentation set.
3912 * Add thread id to rx debugging messages
3914 * When all servers are offline, return a bad network path error
3915 immediately. This is necessary to allow Explorer to do the
3916 right thing and not hang. The background daemon thread will
3917 check the down servers every 30 seconds to see if they are
3920 Since 1.4.0 rc3: 1.4.0 rc4 released (14 Sep 2005)
3922 * The AFSCache file has been modified to store both the
3923 serial number of the volume on which the file resides
3924 and the SID of the local machine. If either change,
3925 a new UUID is generated.
3927 * add "fs flushall" command that forces all data buffers
3930 Since 1.4.0 rc2: 1.4.0 rc3 released (4 Sep 2005)
3932 * add support for '/' instead of '\' in pioctl() calls
3934 * Apply AFS Client Admins group protect to AFS Shell Extension
3936 * Add support for \\afs\<name> to most AFS Shell Extension
3937 commands except the symlink methods.
3939 * when installing the Microsoft Loopback Adapter, enable
3940 MS Client for Networks and prevent an install failure
3941 by not calling CoInitialize twice in the same thread.
3943 * reload cell vldb values from the CellServDB every two
3944 hours in case it changes
3946 * When updating cell information from DNS, be sure to set
3949 * Add support to allow use of \\AFS\<foo> where <foo> is
3950 either a mount point or symlink. As <foo> is normally
3951 treated as a share name, we transform it into \\AFS\all\<foo>
3954 * Init 'code' to prevent false errors when integrated logon
3955 is disabled and the service is not running
3957 Since 1.4.0 rc1: 1.4.0 rc2 released (28 Aug 2005)
3959 * Removed trace log messages that were hampering performance.
3961 * Fixed a deadlock that was being triggered by editing Word
3962 documents stored within AFS with WinWord 2003.
3964 * Bit 3 of the TraceOption registry value is now used to set the
3965 default for "fs trace" in the non-Debug builds. The new default
3966 is off for release builds and on for debug builds.
3968 * Bit 2 of the TraceOption registry value can be used to turn on
3969 real-time output of debug log entries to the Windows Debug
3970 Output monitor interface. This data can be viewed with tools
3971 such as Sysinternal's DbgView.
3973 Since 1.3.87: 1.4.0 rc1 released (17 Aug 2005)
3974 * When the cache manager reports ALLBUSY or ALLOFFLINE for
3975 a resource we should not be returning status codes such
3976 as Network Busy or Remote Host Down because these errors
3977 will imply that the AFS Client Service is Busy or Down
3978 and that is simply not the case. Instead we will return
3979 Bad Network Path as the path is temporarily not available.
3981 Instead of returning Sharing Paused when there is a
3982 timeout error, return Timeout. Once again, the AFS Client
3983 Service is not paused.
3985 * afscreds.exe would display an Obtain Creds dialog when
3986 the expired credentials reminder was triggered even
3987 if there was no network path to the KDC. This is prevented
3988 by adding KDC probe logic to the reminder thread.
3990 * afscreds.exe would display expired tokens no differently
3991 than unexpired ones. This would make it difficult for the
3992 user to distiguish when the tokens were expired. For the
3993 English build added a new resource string "(expired) that
3994 is displayed instead of the expiration time.
3997 * "fs wscell" when executed in freelance mode will return the
3998 name of the cell configured in the registry. The root.afs
3999 volume is not loaded from this cell, but it is used for the
4000 default for aklog and integrated login.
4002 * "fs mkmount, fs rmmount, symlink make, symlink remove" will
4003 no longer work on \\AFS\all when freelance is being used
4004 unless the user is a member of the "AFS Client Admins" Windows
4007 * some more checks performed during persistent cache validation.
4010 * Optimize calls to cm_CheckServer from cm_Analyze. Only check
4011 down servers for the current cell not all cells.
4013 * If the scache LRU list becomes corrupted, the AFS Client Service
4014 might enter an infinite loop at startup. Detect the corruption
4015 and discard the cache contents in this case.
4017 * Fixed the Explorer Shell's remove mountpoint functionality.
4019 * Fixed a deadlock caused by the holder of cm_aclLock attempting to
4020 obtain a mutex lock on a cm_scache_t object whose lock is held by
4021 a thread attempting to obtain the cm_aclLock.
4024 * Really fixed DNS AFSDB queries to ensure that "csail" cannot be
4025 misinterpretted as "csail.mit.edu" when the resolver
4026 is configured to append ".mit.edu" to failed DNS queries.
4028 * Added a new registry key, "LogoffPreserveTokens" (see registry.txt),
4029 that can be used to force the preservation of user tokens upon logout.
4031 * Update the NSIS install scripts to use NSIS 2.07. This release adds
4032 recommended installation categories: AFS Client, AFS Administrator,
4033 AFS Server, AFS Development Kit. Each category includes a different
4034 default subset of the OpenAFS components.
4036 The OpenAFS logo is now associated with the NSIS uninstall entry
4037 in the Add/Remove Programs control panel.
4039 * The user name associated with AFS tokens when obtained with
4040 integrated login, the afs systray tool, or aklog will always
4041 include the full Kerberos 5 user principal name regardless of
4042 whether or not the cell is local to the realm.
4044 * Modify integrated login so that it does not enter an infinite
4045 loop if the service is not set to auto start.
4047 * Added asetkey.exe used to set a Kerberos 5 key for use by the
4050 * Added uninstall.exe to Wix installation
4052 * More modifications to algorithms used to wake sleeping threads.
4056 * Changes to the algorithms used to wake threads when they are
4057 put to sleep because of conflicts over buffer operations.
4060 * The OpenAFS integrated logon obtains Kerberos 5 tickets as part of
4061 the process of producing AFS tokens. The tokens are stored within
4062 the AFS Client Service but the Kerberos 5 tickets have been discarded.
4063 New functionality has been added to temporarily copy the tickets to
4064 a file ccache during the short window between the NPLogonNotify and
4065 WinLogon Logon Event. The Logon Event handler starts a process as
4066 user, afscpcc.exe, that copies the tickets into the default ccache.
4067 The file ccache is then destroyed by the event handler.
4069 * During synchronization operations on stat cache entries, if threads
4070 are forced to sleep it was possible that the threads would not be
4071 woken under heavy load.
4073 * Enforce the delete on close flag specified by the NTCreateX()
4076 * A race condition in the rx library was fixed that resulted in
4077 a deadlock in rx_NewCall when the RX_CONN_MAKECALL_WAITING flag
4078 was cleared when threads were still waiting to use the
4079 connection. Also, fixed a potential case in which the
4080 connection could be destroyed while threads were waiting to use it.
4082 * The DNLC freelist has been seen to become corrupted with still
4083 active entries being stored on the freelist. Changes were made
4084 to perform a better job of cleaning entries before placing them
4085 onto the freelist; marking them with the correct magic value when
4086 purging the list; and allowing the list to be purged on startup
4087 if the cache validation fails.
4089 * Windows Crash Reporting does not get triggered for afsd_service.exe
4090 because it provides its own top level exception handler. This patch
4091 places an explicit call to ReportCrash() on platforms which support
4094 If you configure Crash Reporting via AD Policy to capture crash
4095 reports within the domain, then you will receive this output.
4097 * Generate MiniDumps independent of the Windows Crash Reporting.
4099 (a) If an exception occurs,a minidump will be generated at the path
4100 %TEMP%\afsd.dmp. The type of minidump defaults to Normal but
4101 can be set to any other type via use of the
4104 [HKLM\SOFTWARE\OpenAFS\Client] MiniDumpType (DWORD)
4106 (b) MiniDumps can be generated at any time using the "fs minidump"
4107 command. This allows you to generate a minidump without
4108 stopping the process or even requiring the use of a debugger
4111 * an RX connection which reports bad ticket now treats it the same way
4112 the expired ticket is treated. The ticket is removed from the
4113 connection and a new anonymous connection is established.
4115 * fs memdump now requires that the caller be in the AFS Client Admins
4118 * additional debugging information is provided in cm_Analyze when the
4119 error code is ignored.
4121 * Fix a regression introduced into the pioctl() client code which
4122 will cause a crash in winlogon.exe, explorer.exe and fs.exe if
4123 MIT KFW is not installed. (1.3.82a)
4125 * Remove AFS Gateway option from OpenAFS Control Panel (1.3.82a)
4128 * Fix a race condition in cm_EndCallbackGrantingCall() that
4129 could leave 'nrevp' pointing to freed memory. In the process,
4130 optimize the processing to ensure that the applications monitoring
4131 the status of the stat cache entry are only notified at most
4134 * Conditionalize access to the Explorer Shell AFS->Mount Point
4135 ->Remove menu item. Disable the item if the selected item(s)
4136 are not mount points.
4138 * Activate AFS RPC (RX) Free Packet Queues stored in Thread
4139 Local Storage. This eliminates thread contention on the
4140 global RX free packet queue. This should improve performance
4141 on multi-CPU systems.
4143 * Fix the IP address filtering code to properly load/unload
4144 the IPHelper DLL so that Windows XP/2003 and beyond systems
4145 do not have to manually probe the registry.
4147 * CleanupACLEnt() was not being called consistently with
4148 the cm_scache_t object referred to by the back pointer
4149 mutex locked. This could in very rare conditions lead to
4150 an invalid memory access.
4152 * Added a script command to the msi installer "afs_replace"
4153 which can be used to replace one version of OpenAFS with
4154 another without requiring an uninstall and reinstall if
4155 the installed msi is no longer in its original location.
4157 * Apparently the problem with multi-domain forests with cross-
4158 realm trusts to non-Windows realms was not entirely solved.
4159 The authentication to the AFS SMB service failed because
4160 the wrong name was being used. Using ASU as an example,
4161 the authentication was being performed with the name
4162 "QAAD\user" (an account in the forest root) and not
4163 "user@ASU.EDU (the MIT Kerberos principal used to login with)
4165 The solution was to add an additional dependency on KFW
4166 in order or to be able to easily obtain the client principal
4167 name stored in the MSLSA ccache TGT. This information is
4168 used in two locations:
4170 - the pioctl() function
4172 - a new WinLogon Event Handler for the "logon" event.
4174 The pioctl function will now be able to use the correct
4175 name when calling WNetAddConnection2() and the "logon"
4176 event handler will now be able to call WNetAddConnection2().
4177 The hope is that the "logon" event handler will be called
4178 before the profile is loaded but I have not guarrantee
4182 * Fixed a locking error in cm_TryBulkStat() which had the
4183 potential of deadlocking the system for the length of time
4184 it takes to perform a bulk status fetch operation.
4186 * Replaced time conversion code (UnixTime <-> FILETIME) to
4187 be completely arithmetic instead of relying on a bizarre
4188 algorithm involving a variety of C RTL time functions.
4189 This has the side effect that UnixTime and FILETIME which
4190 are both stored in UTC are interpretted as UTC throughout
4191 the year. Windows will apply the same localization to AFS
4192 as it does to NTFS. Applications which rely on the ability
4193 to sync files between the two file systems will no longer
4194 see the timestamps of files in AFS change an hour relative
4195 to the files stored in NTFS or Windows based backup devices.
4197 * Fixed a invalid memory access under a bizarre circumstance.
4198 Windows will allow a physical mass media device to be
4199 installed via Plug N Play to the system and will assign it
4200 the lowest available drive letter. This is true even if the
4201 drive letter is currently assigned in the user session to
4202 a network device via NET USE (or its equivalent.) When this
4203 happens, queries sent to the CIFS server will contain invalid
4204 data. This invalid data was not being caught by the AFS
4205 Client Service and was resulting in a crash when Freelance
4208 * Fixed a reference count error when registering callbacks
4209 on a stat cache object if the callback was already registered
4211 * Add a case to cm_Analyze() in order to handle the case where
4212 cm_GetConnByMServers() returns CM_ERROR_NOSUCHVOLUME because
4213 the server list for the volume is empty. In this case, force
4214 an update of the volume info and retry.
4216 * Insert a missing cm_EndCallbackGrantingCall() which could
4217 result in threads waiting for a callback to complete to never
4220 * In the persistent cache, there is a maximum number of volume
4221 entries. Allow volume entries to be reused if the maximum
4222 number have been allocated and their reference count is zero.
4224 * If we already have a dead virtual connection object, don't
4225 ignore the fact that additional dead vc objects must be taken
4228 * Removed a deadlock condition introduced in the summer of 2004.
4229 Do not hold mx locks around calls to RX functions including
4230 rx_NewCall(). That is what reference counts on the rx_connection
4233 * Fixed an initialization error in afslogon.dll which could
4234 result in random behavior including a failure to terminate
4235 the NPLogon function. (1.3.80b)
4237 * Fixed an error preventing the use of SMB authentication
4238 on Windows 2000. (1.3.80a)
4241 * Updated CellServDB to Public CellServDB 16 Mar 2005
4243 * Fixed DNS AFSDB queries to ensure that "csail" cannot be
4244 misinterpretted as "csail.mit.edu" when the resolver
4245 is configured to append ".mit.edu" to failed DNS queries.
4247 * Fixed another case in which the client would replace connections
4248 to the host when it was not necessary. In this case the
4249 situation would occur if cryptall was on and the connections
4250 were unauthenticated due to lack of an appropriate token.
4252 * OpenAFS for Windows has failed to work at sites which are
4253 utilizing a cross-realm trust between an MIT/Heimdal realm
4254 and a multi-domain Windows forest when the workstation being
4255 accessed is not located in the root domain. This is caused
4256 by a bug in the workstation which was triggered after the
4257 introduction of Windows 2003 Server. When the bug is triggered,
4258 the workstation attempts to authenticate users to afsd_service.exe
4259 by contacting the Domain Controller instead of using the
4260 LSA loopback authentication mechanism.
4262 One of the reasons this bug occurs is because the workstation
4263 does not have a reliable way of knowing that the service whose
4264 netbios name is "AFS" is located on the workstation. This will
4265 be fixed starting in Longhorn Beta 1 by Microsoft. The
4266 "BackConnectionHostNames" registry value will be used to
4267 indicate that the authentications to that service name should
4268 be performed using the loopback authentication mechanism.
4270 In the meantime, when Logon Caching is enabled, we can force
4271 afsd_service.exe to authenticate using the logon cache before
4272 contacting the Domain Controller. This will work with both
4273 password and smart card based logons.
4275 * The allDown logic in cm_ConnByMServers() was wrong. The allDown
4276 flag should not be cleared if a volume's server reference is
4277 marked as "offline". In the case where all of the volume's
4278 servers are either "down" or the volumes are "offline", we want
4279 cm_Analyze() to process the condition as CM_ERROR_ALLOFFLINE
4280 instead of as CM_ERROR_TIMEDOUT. In fact, CM_ERROR_TIMEDOUT
4281 should never occur in practice.
4283 In the case of CM_ERROR_ALLOFFLINE, cm_Analyze() will sleep for
4284 5 seconds, clear the server down and volume busy flags, and
4285 then force an update from the VLDB. This allows the client to
4286 update the location of a volume if the reason for it being
4287 marked offline is because it is being moved. Calls to
4288 cm_ConnByMServers() will be retried either until success or
4289 the RDRtimeout period is reached.
4291 * Correct the Power Management code behavior in response to
4292 Standby, Suspend, and Shutdown events. Instead of flushing the
4293 buffers associated with the mounted SMB submounts, simply write
4294 all buffers which are marked dirty.
4296 * Added support for "TheseCells" to afslogon.dll. "TheseCells"
4297 provides a list of cells other than the default cell for which
4298 tokens should be obtained using the default Kerberos principal.
4299 This functionality is logon domain specific and is only available
4300 when using KFW for authentication.
4302 * Fixed FindNext Invalid Handle error caused by over agressive
4303 attempts at garbage collecting dirSearch entries when the
4304 dirSearch ID wraps from the maximum value to 1.
4306 * Add support for registry defined server preferences for VLDB and
4307 File servers. See registry.txt for details.
4309 * Increased default cache size to 96MB and default number of cache
4312 * Fixed refCount leaks related to directory lookups and pioctl calls.
4314 * Callbacks revoked during a race condition with an attempt to
4315 obtain the same callback no longer result in an inaccessible
4318 * New command line tool:
4320 afsdacl : Set or reset the DACL to allow starting or stopping
4321 the afsd service by any ordinary user.
4323 Usage : afsdacl [-set | -reset] [-show]
4324 -set : Sets the DACL
4325 -reset : Reset the DACL
4326 -show : Show current DACL (SDSF)
4328 * IP addresses are no longer queried once at startup. Instead IP
4329 addresses are obtained as needed. Loopback adapter addresses are
4330 no longer published to the server.
4332 * Pay attention to the MIT KFW registry configuration for automatic
4333 importation of MSLSA credentials. (SOFTWARE\MIT\Leash32,MsLsaImport)
4335 * Fix (once again) case-sensitive comparisons which was apparently
4338 * Activate support for vos listvol -format
4340 * Population of the "AFS Client Admins" group failed on non-English
4341 versions of Windows because the name of the "Administrators" group
4342 is localized. Now we lookup the name of the group by using the
4343 Administrators Group Alias Relative ID.
4345 * Multiple cell token acquisition within afscreds.exe was broken in the
4346 case of Kerberos 5 cross realm authentication. Instead of contacting
4347 the KDC associated with the cell's realm, afscreds would obtain a
4348 token from the KDC of the user's realm. This would result in
4349 "invalid kvno errors" while authenticating to the AFS servers.
4350 Unauthenticated access would work. The symptoms would vary based upon
4351 whether or not the VLDB servers had been contacted using unauthenticated
4352 connections prior to the user obtaining tokens.
4354 * The list of ACL entries was becoming corrupted. It appears as if
4355 an ACL when it expires was not being appropriately cleaned up. In
4356 fact, it was left in the list of ACLs associated with the scache
4359 * Changed the default @sys name list to "x86_win32 i386_w2k i386_nt40"
4360 for 32-bit x86 systems. The default for itanium will be "ia64_win64"
4361 and "amd64_win64" for amd 64-bit processors.
4363 * When executing executables, Windows provides a mechanisms by which
4364 the normal search for DLLs can be bypassed. If a file foo.exe is
4365 being executed and there exists a file or directory "foo.exe.local"
4366 (the contents are ignored if a file), then Windows will search for
4367 DLLs first in the "foo.exe.local" directory and second in the
4368 directory in which the "foo.exe" file is located.
4370 Previous releases of OAFW would improperly return
4371 STATUS_NOT_A_DIRECTORY instead of the expected
4372 STATUS_OBJECT_PATH_NOT_FOUND. This would cause the Windows SMB client
4373 to terminate the search for the DLL causing the execution of the
4374 application to fail.
4376 * Rework the reference counting of smb_vc_t objects. The references
4377 stored in the waiting locks were not counted. This could result in
4378 an assertion if the reference count drops to zero.
4380 * "fs wscell" returns "Freelance.Local.Root" for the workstation cell
4381 when running in freelance mode.
4383 * Added support for persistent caching of file contents, stat entries,
4384 volume data, cell data, and name lookup data. When the registry
4385 value, "NonPersistentCaching", is not defined or set to 0 the file
4386 specified by "CachePath" becomes a persistent cache file. The
4387 size of the cache file is computed at run time. It has a maximum
4388 size of approximately 1.2GB. The contents of the file will be
4389 validated according to the rules specified by the "ValidateCache"
4392 * Added support for UUIDs. UUIDs are kept for as long as the
4393 cache file is intact.
4395 * Added cmdebug.exe and support cache manager debugging callback
4396 interfaces in afsd_service.exe.
4398 Usage: cmdebug -servers <server machine> [-port <IP port>] [-long]
4399 [-addrs] [-cache] [-help]
4400 Where: -long print all info
4401 -addrs print only host interfaces
4402 -cache print only cache configuration
4404 * Symlinks to \\AFS\[all\]... will now be treated the same as
4405 symlinks to /afs/.... However, please use /afs/... as otherwise
4406 the symlinks won't work on Unix.
4408 * Correct a problem with local CellID allocation for cells whose
4409 server list is obtained via DNS instead of a CellServDB file. If
4410 the DNS information expires the CellID assigned to the entry will
4411 be changed. This causes all of the cm_scache_t objects which refer
4412 to the old cellID number to become useless. Attempts to access
4413 files or directories with cm_scache_t objects using the old cellID
4414 will fail since the server list cannot be obtained.
4416 * Correct deadlock condition in cm_EndCallbackGrantingCall which can
4417 be triggered if the volume referenced by the cm_scache_t object is
4420 * The AFS Service needs to respond to SERVICE_ACCEPT_SHUTDOWN messages
4421 in addition to SERVICE_ACCEPT_STOP.
4423 * Move RPC shutdown until after the SMB and RX shutdown procedures
4424 complete. Block until RPC shutdown is complete.
4426 * Modify afslogon.dll (integrated logon) to wait for service if its
4427 state is START_PENDING. If the timeout period occurs, reset to
4428 the retry interval and not the sleep interval.
4430 * When renewing the server list for a cell obtained via DNS AFSDB
4431 records, the cm_cell_t entry must be removed from the list of all
4432 cells. Otherwise, the list of cells will be corrupted.
4434 * In the dcache and scache modules, use the cm_scache_t dataVersion
4435 instead of the cm_fakeDirVersion.
4437 * Synchronize fs.c with the unix version.
4439 * The variable used to determine whether a file or virtual memory
4440 mapped cache is used was not properly initialized to a default
4441 value. If the registry setting "NonPersistentCaching" was not
4442 set, the choice would be random. Properly initialized to be
4445 * The memory mapped view was never unmapped before closing the file
4446 at service shutdown. This is now properly cleaned up.
4448 * Default location of Cache file is now %TEMP%\AFSCache
4452 * A bug affecting new installations of 1.3.75/76 would result in
4453 the creating of incorrect mountpoints in the freelance root.afs
4454 volume for the default cell.
4456 If "fs \\afs\all\<cellname>" lists a volume name of "root" instead
4457 of "root.cell", you have been affected by the bug. To correct the
4458 problem, execute the following commands:
4460 fs rmmount \\afs\all\<cellname>
4461 fs rmmount \\afs\all\.<cellname>
4463 fs mkmount \\afs\all\<cellname> root.cell <cellname>
4464 fs mkmount \\afs\all\.<cellname> root.cell <cellname> -rw
4468 * A bug has been fixed in the auto-generation of Freelance root.afs
4469 symlinks which produced random entries in the \\AFS\all directory.
4471 * Support has been added for multi-homed servers
4474 * Added a new registry value, "StoreAnsiFilenames", which can be used
4475 to force the use of ANSI character sets instead of OEM Code Pages.
4476 This feature is useful when users require the ability to create
4477 filenames with 8-bit characters and need to access the files from
4478 both Latin-1 based Unix systems as well as from Windows.
4480 Activation of this feature will prevent access to files stored with
4481 8-bit OEM characters.
4483 * Shutdown all SMB threads in a synchronized manner when stopping the
4486 * There is currently a maximum cache size of 1.3GB. The limit is imposed
4487 by the largest contiguous block of unused memory within the 2GB process
4488 space which can be assigned to the memory mapped file. Unfortunately,
4489 when the executable digital signature verification code is activated
4490 Windows sees fit to further segment the process memory which in turn
4491 reduces the size of the maximum cache file to less then 800MB. If
4492 larger cache sizes are desired, a new registry value should be set:
4494 HKLM\SOFTWARE\OpenAFS\Client (DWORD) "VerifyServiceSignature" = 0x0
4496 Setting this value will disable the runtime verification of digital
4497 signatures on afsd_service.exe and the afs dlls which it loads. It
4498 will not disable the the version number check on those same files.
4499 The signature verification is not a security messure and is only meant
4500 to enhance the ability to afsd_service.exe to detect potential
4501 destablizing mixtures of DLLs from incompatible distributions.
4503 Added code to auto-disable the signature verification check if
4504 the desired cache size is greater then 700MB.
4506 * Windows' WinTrustVerify(WIN_SPUB_ACTION_PUBLISHED_SOFTWARE) is
4507 used to verify the validity of the afsd_service.exe binary
4508 as well as each of the AFS DLLs loaded by the service. Not only
4509 must the digital signature be valid but the signatures of the
4510 DLL must be signed by the same entity as the service.
4512 * Implement new functions: cm_freelanceMountPointExists and
4513 cm_freelanceSymlinkExists. Use them along with other validity
4514 checks in cm_freelanceAddMount and cm_freelanceAddSymlink to
4515 ensure that name collisions do not occur and that empty strings
4516 are not valid file names.
4518 A symlink may not have a name which would resolve to a valid
4519 cell name. Doing so would prevent access to the cell.
4521 * Add missing cm_HoldSCacheNoLock call to Freelance mount point
4522 re-initialization code. The reference counts of the fake root.afs
4523 volume scache object(s) would become invalid when the mount point
4524 or symlink lists were altered.
4526 * Add registry entries to provide mappings from the afsdsbmt.ini
4527 to the new locations for applications which count on the use
4528 of the old Profile file APIs. These apps are likely to fail
4529 if the user does not have administrator privileges and the
4530 registry is locked down.
4532 * The afs_config.exe submounts dialog had two errors.
4533 First, attempts to remove entries failed because the registry
4534 key was being opened without KEY_WRITE privileges.
4535 Second, when editing a submount entry, changing the name
4536 would add a new key and leave the original one in place.
4537 Now the original submount will be removed if its name is
4540 * In recent months there have been several incidents in which
4541 users have experienced problems starting or accessing
4542 afsd_service.exe and after significant effort has been spent
4543 it has turned out that they have two versions of AFS on the
4544 machine or an inconsistent set of DLLs.
4546 Code has now been added to afsd_service.exe which will walk
4547 the list of modules loaded by afsd_service.exe and validate
4548 that the version of the AFS DLLs matches the version of the
4549 afsd_service.exe executable. If they do not match the service
4552 * When Freelance mode is enabled and there is no registry
4553 key HKLM\SOFTWARE\OpenAFS\Client\Freelance, afsd_service.exe
4554 will attempt to import the afs_freelance.ini file contents.
4555 If the file does not exist, it was creating a dummy file
4556 with a r/o and r/w entry for the default cell and then
4557 importing those values.
4559 This process has been changed. The temporary file is no
4560 longer created. Also, both the OpenAFS Client install
4561 directory as well as %WINDIR% are checked for previous
4562 afs_freelance.ini files.
4564 * Added support for VL_GetEntryByNameN(). Still need to add
4565 support for VL_GetEntryByNameU() for multi-homed support.
4567 * Fix a deadlock situation in afscreds.exe when canceling an
4568 auto-generated Obtain Tokens dialog
4571 * Fix the locking of objects during Directory Searches in the
4572 SMB/CIFS server. The failure to properly lock the reference
4573 counts was resulting in the premature freeing of smb_dirSearch_t
4574 objects while they were still in use by the SMB/CIFS client.
4575 This does not solve the "Invalid Handle" problem.
4577 * Fix Find Cell By Name pioctl call to return a valid cell
4578 name for the Freelance fake root.afs volume.
4579 "Freelance.Local.Root".
4581 * Fix the Explorer Shell Extension Symlinks->Add operation.
4582 The dialog template was missing and the link destination
4583 string was too short.
4585 * Add support for symlinks to Freelance root.afs volume
4586 Stored at HKLM\SOFTWARE\OpenAFS\Client\Freelance\Symlinks
4587 <number> = "<linkname>:<relative-path>."
4588 Use symlink.exe to create, list, or remove
4590 * Remove the fallback to the use of KFW's KRB4 library when obtaining
4591 tokens. We never obtain KRB4 tickets.
4593 * Fix AFS Client Configuration Control Panel to support new SysName
4596 * Fix a bug in afsd_service.exe which could result in the SysName
4597 not being read from the registry.
4600 * Add code to block the issuance of AFS tokens by aklog.exe or
4601 afscreds.exe when the Kerberos 5 principal name contains a dot.
4603 * Modify the IsAdmin() function to always treat the local SYSTEM
4604 account as an AFS client administrator. Affects fs.exe and
4607 * Modify the internal handling of Quota Exceeded errors
4609 * Upgrade all reference count fields in the Windows cache manager
4610 and the osi library to use unsigned long instead of signed short.
4611 A similar fix has been applied to the afs rpc (rx) library.
4613 * fix the Windows cache manager to prevent it from replacing the
4614 rx_connection object associated with the cm_conn_t object on each
4615 and every operation if "fs crypt" was set. This explains the
4616 dramatic performance difference when crypt is used vs clear.
4617 The problem: 'cryptall', a boolean flag indicating whether or not
4618 "fs crypt" is set, was being compared to the rx_connection
4619 cryptlevel which is either rxkad_clear:0 or rxkad_crypt:2.
4620 1 != 2 and therefore the rx_connection was always destroyed
4621 and replaced on each and every operation.
4623 Lock the cm_conn_t object around every call to RXAFS_xxxx functions.
4624 It is not safe for the cm_conn_t object to not be locked because
4625 rx_DestroyConnection might be called from another thread if:
4626 - the user's tokens have changed or expired
4627 - the crypt mode has changed
4629 This fix appears to have also taken care of the problems associated
4630 with Overlapped Writes resulting in Delayed Write errors.
4632 * fix NSIS installer's AdminGroup.exe to properly create and
4633 remove groups when given -create or -remove. The string comparison
4636 * fs sysname now accepts a list of sysname values
4638 * added a new registry value HKLM\SOFTWARE\OpenAFS\Client "IoctlDebug"
4639 DWORD which when set to a non-zero value will cause error message
4640 text to be output to stderr from the pioctl() routine. Useful in
4641 debugging failures of fs.exe, tokens.exe, etc.
4643 * added a test to the power management code to only perform a
4644 flush operation if there is at least one network adapter which
4645 is not a loopback adapter.
4647 * Fix bug in loading of registry value HKLM\SOFTWARE\OpenAFS\Client
4648 "EnableKFW". This value will not be read if the key
4649 HKCU\SOFTWARE\OpenAFS\Client exists; even if the "EnableKFW"
4650 value under that key does not.
4652 * provide mechanisms to force the use of krb524d for Kerberos 5
4653 ticket to AFS token conversion. For afslogon.dll and afscreds.exe
4654 there is a new registry value "Use524" and for aklog.exe a new
4655 command line parameter "-m".
4657 * Fix the pattern matching algorithm to properly match patterns
4660 * smb_ReceiveCoreRename() was factored to produce smb_Rename()
4661 which is used by both the original function and the new
4662 smb_ReceiveNTRename(). smb_ReceiveNTRename() supports the
4663 creation of HardLinks in addition to Renaming. smb_Link()
4664 is a new function which creates HardLinks via cm_Link().
4665 cm_Link() is a new vnodeops function which creates links
4668 smb_ReceiveNTRename() does not support the File Copy and
4669 Move Cluster Information operations described in its interface.
4670 ReceiveNTRename is under documented in CIFS-TR-1p00_FINAL.pdf.
4672 * When opening files via symlinks, we should follow the symlinks
4673 until we reach the actual file stat cache entry. The stat cache
4674 entry of the file should then be stored in the FID instead of
4675 stat scache entry of the symlink.
4677 * return bad operation errors for all unimplemented functions
4678 even if we do not know the functions exist.
4680 * Log bad packets and unknown operation packets to the trace log
4682 * Map CM_ERROR_BADOP to STATUS_NOT_SUPPORTED instead of
4685 * Update list of known CIFS operations to include all those listed
4686 in CIFS-TR-1p00_FINAL.pdf.
4688 * Modify the handling of HKLM\SOFTWARE\OpenAFS\Client\Submounts
4689 to support the REG_EXPAND_SZ type.
4692 * A new Windows authorization group "AFS Client Admins" is now
4693 created and populated with the members of the "Administrators"
4694 group. The group is used to determine which accounts on the
4695 machine may be used to modify the AFS Client Configuration via
4696 the UI and command line tools. afs_config.exe, fs.exe,
4698 * Modify the WinLogon Logoff Event Handler to query NT4 domain
4699 controllers for the remote profile path if Active Directory
4700 services are not available.
4702 * Fix aklog.exe to not add the AFS ID to the username
4704 * PTS registration of new users to foreign cells has been added to
4707 * The cm_Daemon thread is used to perform checks for
4708 down servers, up servers, volumes, callback expirations,
4709 lock maintenance and token expiration. Due to a gaff in
4710 larger integer division the thread never performed any
4711 work. Instead the current time computation would always
4712 be less then the trigger times. This had an adverse affect
4713 on the client's ability to maintain communication with servers,
4714 keep volumes up to date, and flush user tokens and acls
4715 when they have expired. This was broken when the 1.3 branch
4716 was modified to support VC7 which no longer included
4719 * An initialization problem with the Freelance code was
4720 detected while fixing the callbackRequest. The cm_rootSCachep
4721 object is obtained during afsd_InitDaemons() but the callback
4722 information is incomplete. The callback information will not
4723 be obtained until cm_MergeStatus is called from within
4724 cm_GetCallback. Unfortunately, cm_SyncOp did not properly
4725 test for the conditions under which the callback information
4728 * Reports have been filed indicating that callbacks were
4729 being lost. An examination of the code indicated that the
4730 cm_server_t objects were not being properly reference
4731 counted by the cm_scache_t and cm_callbackRequest_t objects.
4732 In particular, the cm_server_t objects may have been freed
4733 from beneath the cm_conn_t objects.
4735 All of the reference counting is now done via the functions:
4740 this improves the ability to track the referrals.
4742 Each cm_BeginCallbackGranting Call now allocates a reference
4743 to the cm_server_t. The cm_EndCallbackGrantingCall either
4744 frees the reference or transfers it to the cm_scache_t
4745 cbServerp field. These are then appropriately tracked
4746 through the cm_Analyze call.
4748 * Ensure that the dnlc hash table is the same size as the
4749 dir name hash table (as per original author's note).
4750 Increase the dnlc CM_AFSNCNAMESIZE to a multiple of 8
4751 for compatibility with 64-bit systems.
4753 * fix smb_ApplyV3DirListPatches to properly apply the hidden
4754 attribute to dotfiles when the infoLevel < 0x101 and
4755 cm_SyncOp has failed.
4757 * Fix the Freelance registry initialization code. There
4758 was a possibility that some systems could end up with
4759 garbage in the registry during a clean install.
4762 * file and directory names beginning with "." will now be given the
4763 hidden attribute when the volume access is anonymous. this matches
4764 the behavior when the volume access is via an authenticated user.
4766 * Added a change monitor to the HKLM\SOFTWARE\OpenAFS\Client\Freelance
4767 key. When a change occurs mark the root.afs data as invalid and
4768 for it to be reloaded on the next access. This allows administrators
4769 to modify the mount point list without restarting the service.
4771 The freelance client used to provide a fake modification time for
4772 the root.afs volume data and its mount points of 7/09/2001 14:24 EDT.
4773 Added code to extract the last modification time of the Freelance
4774 registry key and use that instead. The time now represents the
4775 most recent mount point change.
4777 * PTS registration of new users to foreign cells has been added to
4780 * Additional Cache Control and Credential Manager options have been
4781 added to the WiX installer. See deployment guide for details.
4783 * The CachePath setting is now optionally a REG_EXPAND_SZ type
4785 * The WiX installer has been upgraded. Version 2.0.1927.1 is now
4788 * The loopback installation code may have had a problem updating the
4789 %ETC%\HOSTS file which could have resulted in a premature failure.
4790 Work around code has been added for the case where the file cannot
4793 * The default max chunksize was increased from 15 (32K) to 17 (128K)
4794 because Windows sends 64K blocks when using overlapped writes.
4796 * The default number of server threads was increased from 4 to 25 to
4797 better handle overlapped writes.
4799 * The "AfscredsShortcutParams" registry value was not being properly
4800 loaded by afscreds.exe. Therefore, the default value was always being
4801 used instead of the value set by the installer.
4803 * Windows XP provides downgrade attack detection to prevent an attacker
4804 from being able to force the use of NTLM simply by disrupting
4805 communication with the KDC. This attack cannot exist between the
4806 Windows CIFS client and the AFS Client Service. Therefore, when a
4807 downgrade has been detected the afs pioctl library will force the
4808 establishment of a new CIFS connection using NTLM.
4810 * A locking error was discovered surrounding all references to volume
4811 server lists within the cm_cell.c source file.
4813 * The logged into Windows username was incorrect on Terminal Server
4816 * A new registry value "NonPersistentCaching" was added to the service
4817 parameters key. When set to a non-zero value, the afs cache is stored
4818 in the Windows paging file. There are two limitations to choosing
4820 1. when persistent caching is implemented it won't work with
4821 this flag set since there will be nothing to persist.
4822 2. with this flag set the initial paging allocation cannot be
4823 changed while the service is running
4825 * An initialization bug was discovered in aklog.exe which affected users
4826 who have a domain name for their afs servers which could not be mapped
4830 * afs_config.exe now validates cell names against DNS in addition
4831 to the CellServDB file.
4833 * In order to allow the freelance client to connect to a volume with ID
4834 equal to 1 on the default cell we changed the fake root.afs volume ID
4835 once again. This time we choose 0xFFFFFFFF. In addition, we change
4836 the cell ID of the fake root.afs volume from 1 to 0xFFFFFFFF as well.
4837 It will now be impossible for a volume ID to match that of another
4838 cell unless the client is connected to 0xFFFFFFFD cells. That should
4839 be enough room for growth.
4841 * Fix "fs mkmount" command to work with UNC paths and when
4842 started from non-AFS drives. It is now possible to create a mount
4843 point in the freelance fake root.afs volume with the command
4845 fs mkmount \\AFS\all\<directory-name> <volume-name> <cellname>
4849 fs mkmount \\AFS\all\openafs.org root.cell openafs.org
4850 fs mkmount \\AFS\all\.openafs.org root.cell openafs.org -rw
4852 * The algorithm used to re-attempt access to the servers associated with
4853 a volume has been altered to properly address the case in which all
4854 servers have been marked down. The previous algorithm did not reset
4855 the server's down flags so the servers were never actually retried.
4856 This caused a problem with active volumes if the network connectivity
4857 was lost as could be the case with a network cable removal, wireless
4858 drop, or laptop hibernation. With the fix volume access is restored
4859 almost instantenously when network connectivity becomes available.
4861 * Support for SMB/CIFS browsing has been added to the AFS Client Service
4862 SMB server. It is now possible to use "NET VIEW \\AFS" to obtain a
4863 listing of AFS submounts and freelance mount points. Support for
4864 NETSHAREENUM, NETSHAREGETINFO, NETSERVERENUM2, NETSERVERGETINFO
4865 significantly enhances the behavior of AFS volumes within the Explorer
4866 Shell. For instance, "AFS" now shows up as server in the Explorer
4867 with each submount or freelance mount point visible as a share.
4868 The right click menu in each folder now works with full functionality
4869 on a consistent basis.
4871 * The network provider can be configured to have different behavior
4872 depending on the domain that the user logs into. These settings are
4873 only relevant when using integrated login. A domain refers to an
4874 Active Directory (AD) domain, a trusted Kerberos (non-AD) realm or the
4875 local machine (i.e. local account logins). The domain name that is
4876 used for selecting the domain would be the domain that is passed into
4877 the NPLogonNotify function of the network provider. (see registry.txt
4880 * Added a new registry value [HKCU\SOFTWARE\OpenAFS\Client]
4881 "Authentication Cell" which may be used to specify a default
4882 authentication cell for afscreds.exe which is different from
4883 the default cell for the AFS Client Service daemon.
4885 * Added a Logoff WinLogon Event Notification function to afslogon.dll.
4886 afslogon.dll moved to %WINDIR%\System32\.
4887 New registry entries added to register the dll for Winlogon events.
4889 The logoff event will now force a call to ktc_ForgetAllTokens()
4890 using the context of the user being logged off as long as the
4891 user's profile is not loaded from within AFS. If the profile
4892 was loaded from AFS we can't release the tokens since the Logoff
4893 event is triggered prior to the profile being written back to
4894 the its source location. This is now performed in an XP SP2
4897 * Windows XP SP2 Internet Connection Firewall interoperability
4900 * The %WINDIR%\afsdsbmt.ini contains four sections:
4901 Submounts, Drive Mappings, Active Maps and CSC Policies.
4902 The Submounts and CSC policies are now stored in the registry under
4903 [HKLM\SOFTWARE\OpenAFS\Client\Submounts]
4904 [HKLM\SOFTWARE\OpenAFS\Client\CSCPolicy]
4905 The Drive Mappings and Active Maps are stored in the registry under
4906 [HKCU\SOFTWARE\OpenAFS\Client\Mappings]
4907 [HKCU\SOFTWARE\OpenAFS\Client\Active Maps]
4909 There is no automatic migration of this data as it would be impossible
4910 to consistently migrate data to user profiles which may not be active
4911 when the machine is updated.
4913 * The %WINDIR%\afs_freelance.ini contains lists of mountpoints for the
4914 fake root.afs volume. For the same reasons as for the cellservdb file,
4915 this information should not be in %WINDIR%. This information is now
4916 kept under the registry key
4917 [HKLM\SOFTWARE\OpenAFS\Client\Freelance]
4919 The data from the afs_freelance.ini file will be automatically
4920 migrated to the registry on first execution of afsd_service.exe
4922 * Keeping the CellServDB file in the location %WINDIR%\afsdcell.ini is
4923 troublesome for several reasons. One, it is confusing for those who
4924 expect the file to be named "CellServDB" instead of "afsdcell.ini".
4925 Two, this file is not a Windows Profile formatted file. Three,
4926 applications should not be reading or writing to %WINDIR%. It causes
4927 problems for Windows Terminal Server.
4929 The new location of CellServDB will be the OpenAFS Client install
4930 directory which is by default C:\Program Files\OpenAFS\Client and can
4931 be determined by querying the registry for
4932 [HKLM\SOFTWARE\TransarcCorporation\AFS Client\CurrentVersion]PathName
4934 The existing afsdcell.ini will be migrated by the NSIS installer.
4935 The Wix installer must still be updated to do the same.
4937 * Change NSIS installer to use DNS by default; to remove Integrated Logon
4938 High Security mode; and to add Terminal Services compatibility registry
4939 entries to allow the OpenAFS tools to find the afsdcell.ini and other
4940 configuration files in %WINDIR%.
4942 * Add support for authenticated SMB connections. This will remove
4943 the need for high security mode in most situations. Both NTLM
4944 and Extended Security (GSS SPNEGO) modes are supported. Effectively,
4945 only NTLM can be used even though Kerberos is now supported. The
4946 reason is that it is not possible to construct a service principal
4947 which is unique to each individual machine.
4949 SMB Extended Auth does not work on XP SP2 unless one of two registry
4950 modifications are made:
4952 (1) To disable the check for matching host names on loopback connections
4953 set this key. This does not require a reboot:
4955 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa]
4956 "DisableLoopbackCheck"=dword:00000001
4958 (2) To add the AFS SMB/CIFS service name to an approved list. This
4959 does require a reboot:
4961 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa\MSV1_0]
4962 "BackConnectionHostNames"=multi-sz "AFS" "MACHINE-AFS"
4964 afsd_service.exe will automatically add the current Netbios Name
4965 to the BackConnectionHostNames list and then temporarily disable
4966 the loopback check for one cycle of startup/shutdown of the service.
4967 We assume most folks do not start/stop without a reboot so this
4968 will be adequate in most cases.
4970 * Fix security hole in afslogon.dll which allowed passwords to be
4971 sent in clear text to the KDC in a misformed principal name.
4973 * Fix cm_GetCell() to properly handle expired dns entries
4976 * If Freelance mode is active and the afs_freelance.ini
4977 file does not exist, do not create an empty file.
4978 Instead create a file containing ro and rw mountpoints
4979 to the default cell using the standard conventions.
4981 * Modify the Freelance support to handle the ability
4982 to create rw mount points in the fake root.afs volume.
4984 * Changed the RPC mechanism used for token setting from
4985 named pipes to local. Use of named pipes can be restored
4986 by setting the environment variable AFS_RPC_PROTSEQ to
4989 Named pipes were required when a Windows 9x system was
4990 using a NT system in gateway mode which is incompatible
4991 with our use of local loopback adapters.
4993 * In afscreds.exe, if a username of the form user@REALM is
4994 specified and no password is specified, do not perform a
4995 kinit operation. Only perform the aklog functionality.
4997 * Add a new registry value which allows the number of processors
4998 on which afsd_service.exe executes to be restricted. Valid
4999 values are 1..numOfProcessors
5001 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
5005 * A second MSI based installer option is now available.
5007 * Fixed Kerberos 5 kinit functionality in afscreds.exe to properly
5008 request tickets for user/instance@REALM instead of just user@REALM
5010 * Modify the Power Management Notify routine to wait for the Hard Dead
5011 timeout period instead of a fixed 19 seconds. With the longer timeout
5012 periods Hibernation and Standby could never succeed when network
5013 connectivity is not available.
5015 * The following fs.exe commands are now restricted to Administrator:
5016 - checkservers with a non-zero timer value
5019 - sysname with a new sysname list
5028 setting the default sysname for a machine should be done via the
5029 registry and not via "fs sysname".
5031 * NSIS installer adds options to install Debugging Symbols
5032 and the Microsoft Loopback Adapter; the user is now also
5033 given the ability to select the afscreds.exe startup options.
5035 * Build system modified to generate symbols for FREE (aka RELEASE)
5036 builds as well as CHECKED (aka DEBUG) builds
5038 * Sites which have a volume ID of 0x20000001 assigned to their
5039 root.cell volumes have been experiencing problems with accessing
5040 the root.cell volume of their cell when Freelance mode has been
5041 active. This was because 0x20000001 was assigned to the fake
5042 root.afs volume created by freelance. The fake volume id is
5043 now set to 0x00000001 to prevent conflicts.
5045 * The timeout logic in the AFS Client Service has been wrong
5046 for sometime. It is based on two different assumptions.
5047 First, the SMB client timeout is a fix value as was the case
5048 with OS/2 Lan Manager. This assumption is incorrect. The
5049 SMB timeout in Windows is a dynamic value computed based upon
5050 a fixed minimum timeout to which is added time based upon the
5051 size of the request and the performance characteristics of
5052 the connection. Second, it is the responsibility of the
5053 SMB Server to enforce the timeout requirements of the client.
5054 This is untrue. The SMB Server cannot be expected to know
5055 the requirements of the client. More importantly, if the
5056 SMB server uses the SMB client timeout as a value to restrict
5057 its behavior as an RX client, the performance characteristics
5058 of the local SMB session would be used to prematurely terminate
5059 WAN connections with significantly different performance
5062 The timeout logic has therefore been modified in the following
5064 . the Lan Manager Workstation (SMB) Session Timeout is used only
5065 as a basis for configuring the Connection Dead Timeout
5066 and Hard Dead Timeout values. The Connection Dead Timeout
5067 must be at least 15 seconds longer than the SMB Timeout
5068 and the Hard Dead Timeout must be at least double the
5069 Connection Dead Timeout.
5070 . New registry entries have been added to allow the Connection
5071 Dead Timeout and Hard Dead Timeout values independent of the
5072 Lan Manager Workstation Session Timeout
5073 . The test to enforce the SMB Client Timeout has been removed.
5075 One of the side-effects of removing the enforcement of the SMB
5076 Client Timeout is that regardless of whether or not the SMB client
5077 is available to receive the response (and how would the SMB server
5078 know) the RX protocol response can be used to update the AFS
5079 Client Service state for ready access by future SMB client
5082 This should be the end of the "Server paused or restarting messages"
5084 * Add support for arbitrary UNC paths to the pioctl() support.
5085 This enables the fs commands as well as the AFS Shell Extension
5086 to work correctly when UNC paths are being used.
5088 * Fix afscreds.exe (by updating afskfw.lib) to search for cells via
5089 DNS if the cell configuration cannot be determined via CellServDB
5091 * Add debug info to test whether CM_BUF_WAITING or CM_SCACHE_WAITING
5092 are ever set more than once at a time
5094 * Fix the management of lists of cm_cell_t structures when using
5095 DNS to lookup cell information. The previous code would fail to
5096 reuse the same cellID for a cell if DNS was used more than once
5097 for a given cell name. When the ttl expired, a single cm_cell_t
5098 could be inserted into the cm_allCellsp list more than once
5099 producing a loop. In addition, the vlServerp list belonging to
5100 the cell was not freed resulting in improper refCounting of the
5103 * Add DNS support to cm_IoctlNewCell() which previous only examined
5106 * Add cm_FreeServer() function and call it from cm_FreeServerList()
5107 to properly garbage collect cm_server_t objects
5109 * Add numVCs variable to smb.c to track the number of smb_vc_t
5110 objects created and use it to initialize the vcID field which
5111 previously was set to 0 in all objects resulting in FindByID
5114 * Fixed DNS lookups to work consistently throughout the OpenAFS
5115 product instead of just from within the afsd_service.exe
5117 * Added a runtime check to ensure that AFS Client Service SMB
5118 Server is accepting connections before attempting to mount
5121 * Read IP addresses for volume servers out of the CellServDB
5122 file if gethostbyname() on the hostname fails.
5124 * Fix getcellconfig() to populate both the Hostnames as well
5125 as IP addresses when loading cell data via DNS
5127 * Increase the Connection Dead Time to 50 from 20 seconds
5128 Increase the Hard Dead Time to 120 from 40 seconds
5129 (matches the Unix values)
5131 * Fixed an assertion validating the number of allocated NCBs
5133 * Fixed the build environment to consistently build for
5134 Windows 2000 and above. (APPVER = 5.0)
5136 * Fixed rx_debug to properly validate the receipt in incoming
5137 data with select() and recvfrom(). Do not copy data out of
5138 the socket buffer unless success is indicated.
5141 * afsd_service.exe will now display a message box to the
5142 desktop when it terminates due to an IP Address Change.
5144 * installer no longer deletes AFS Server configuration data
5147 * installer generates a warning dialog if the RPC service
5148 is not properly configured
5150 * installer compressed with lzma instead of bzip2
5152 * afsd_service.exe shutdown crash solved once and for all
5154 * reference counting of smb_vc_t data structures improved
5156 * name space collision of smb_fid_t event objects corrected
5158 * the output of "fs memdump" is now written to
5159 %WINDIR%\TEMP\afsd_alloc.log
5161 * the file TaAfsApp_1033.dll is now properly installed allowing
5162 the User Manager to start
5164 * a new algorithm is used for computing filename pattern matches
5166 * afscreds.exe now accepts user names containing instance
5169 * Fix the Directory Name Lookup Cache to be case-sensitive.
5170 This is crucial in environments in which a Windows client
5171 is accessing a directory with more than one filename that
5172 differs only by case. If the directory contains "FOO"
5173 and "Foo". You want "DEL Foo" to delete the correct one.
5174 We still have a problem in that "DEL foo" will delete a
5175 random filename. This will be addressed in a future release.
5177 * Fix afscreds.exe -M option (renewMaps) to work when High
5178 Security mode is off. Also, remember to disable the ActiveMap
5179 flag in afsdsbmt.ini when a drive mapping is removed.
5181 * Updates to NSIS installer script. AFS Server configuration
5182 data will not be destroyed on un-install or re-install.
5183 Use a better compression algorithm.
5185 * afslogon.dll now uses KFW to obtain tokens when available
5187 * afslogon.dll when given an all uppercase username will
5188 attempt to authenticate with both the uppercase name
5189 and an all lowercase variation
5191 * DST modification removed. The fix appears to make things
5192 worse after a reboot of the machine.
5194 * fs.exe: added "cscpolicy" which is used to
5195 change client side caching policy for AFS shares
5197 Usage: fs cscpolicy [-share <AFS share>] [-manual] [-programs]
5198 [-documents] [-disable] [-help]
5200 * Several uninitialized variables have been initialized
5202 * It is now possible to obtain tokens using cross realm
5203 Kerberos within afscreds.exe:
5205 user: jaltman@ATHENA.MIT.EDU
5207 Will obtain a cross realm ticket for jaltman/DEMENTIA.ORG@ATHENA.MIT.EDU
5208 will will in turn be used to obtain afs@DEMENTIA.ORG.
5209 The resulting token will be stored with the display name
5210 jaltman@ATHENA.MIT.EDU@dementia.org
5212 * aklog.exe has been added to the client
5214 Usage: aklog [-d] [[-cell | -c] cell [-k krb_realm]]
5215 [[-p | -path] pathname]
5219 -d gives debugging information.
5220 krb_realm is the kerberos realm of a cell.
5221 pathname is the name of a directory to which you wish to authenticate.
5222 -noprdb means don't try to determine AFS ID.
5223 -5 or -4 selects whether to use Kerberos V or Kerberos IV.
5224 (default is Kerberos V)
5225 No commandline arguments means authenticate to the local cell.
5228 * All of the resource files have been restructured to adhere to
5229 a set of rules IBM implemented for loading string resources.
5230 These rules had either been forgotten or were not discovered
5231 by folks working on the OpenAFS sources. The end result was
5232 memory corruption. This is primary item which was preventing
5233 the AFS Server from working.
5235 * Increased the size of the maximum ticket size stored in a token
5236 from 344 bytes to 12,000. Increased the buffers used to convey
5237 messages between the pioctl() caller and the SMB Server from
5238 1000 bytes to 12,512. The code appeared to have been writing
5239 above the top of the stack by quite a few number of bytes.
5240 (The increased ticket size is necessary for the next item.)
5242 * When obtaining AFS Tokens via KFW, krb524 is no longer required.
5243 Instead the raw Kerberos 5 ticket is used in its entirety. This
5244 is extremely important as it allows us to use pure Kerberos 5 KDCs
5245 as the source of the AFS authentication. The use of up to 12,000 byte
5246 tickets will allow tickets produced by all versions of Microsoft
5247 Active Directory to be used.
5248 - create a user account.
5249 - designate it DES only
5251 - specify its UPN to be "afs@realm"
5252 - assign a SPN of "afs/cellname" to the UPN with setspn.exe
5254 * Do not enforce the funky 8dot3 pattern matching rule that the first "."
5255 is special when using long file names. (you must use "*.*" and not "*")
5256 Instead only enforce it when performing 8dot3 searches.
5258 * Fixed the DST problem with creation times being set one hour ahead
5260 * Fixed the problem when using \\afs\cell-alias. For example,
5261 \\afs\uncc instead of \\afs\uncc.edu. Do not a new cell struct
5262 for the alias name; instead simply expand the name. One of the
5263 symptoms of this problem was a loss of acquired tokens.
5265 * Fixed the AFS Shell Extension. The Symbolic Link menu was empty
5266 of strings. (Only English strings provided.)
5268 * Fixed the installer to properly replace in use files.
5270 * Fixed the build system to cleanup generated component version files
5272 * The release build compiled with MSVC 6.0 compiler to avoid the
5273 afsd_service.exe shutdown crash. This does not solve the problem
5274 but simply avoids it for the time being.
5278 * fix afslogon.dll to not corrupt memory when High Security mode
5281 * fix afsd_service.exe to not attempt to restore the stack when
5282 an exception occurs. (not safe in multi-threaded programs)
5284 * fix uninstaller to properly remove the CRT and MFC DLLs
5286 * remove a Message Box from afscreds.exe when getcellconfig()
5287 fails on a kerberos realm which is not a cell
5289 The following is a list of changes to the OpenAFS for Window client
5292 * "fs setserverprefs" will leave afsd service deadlocked
5294 * "vos listaddrs" will core dump
5296 * installer sets the appropriate keys to support Integrated Logon
5298 * installer disables the "Find Lana by Name" functionality as it
5299 was causing headaches for many users
5301 * fix the intermittent crash of the power management thread when
5302 shutting down the AFS Client Service
5304 * optimizes the obtain drive mount list functionality which is
5305 executed every time the mount tab in afscreds.exe and afs_config.exe
5306 are refreshed. (this happens a lot)
5308 * fix the service shutdown logic. add the STOP_PENDING state
5309 and do not accept additional service events after we declare
5312 The following is a list of changes to the OpenAFS for Window client
5315 * flexelint was run against the source tree and hundreds (perhaps
5316 thousands) of corrections were applied to ensure prototypes
5317 were in use; types were used consistently; variables were
5318 initialized; unused variables were removed; etc.
5320 * A wide variety of instrumentation was added including the
5321 ability to produce a stack trace from within afsd_service.exe
5324 * Dynamic configuration of the RDRtimeout value based upon the
5325 LanMan Workstation Session Timeout
5327 * The mount root no longer needs to be called "/afs". This
5328 is now set by a registry value "MountRoot" within the key
5329 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5331 * The cell list is now only read out of afsdcell.ini when the
5332 file changes instead of each time a cell is resolved.
5334 * Thread synchronization was added to cm_server.c and ktc_nt.c
5336 * All calls to GlobalAlloc()/GlobalFree() were replaced with
5337 calloc()/free(). The Global functions were needed on Windows 3.x
5338 but have caused a variety of problems on the Win32 platforms.
5339 Avoiding them is highly recommended by several Microsoft
5340 Knowledgebase articles
5342 * Support for Symbolic Links added to the AFS Shell Extension
5344 * Added a registry value "OverlayEnabled" to determine if
5345 Shell Extension Overlays should be enabled.
5346 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5348 * New Build system to support VC6, VC.NET, VC.NET2003 compilers and
5349 separate trees for checked and free builds. Build system supports
5350 a custom directory src\WINNT\extra which can be used as a grafting
5351 location of organization specific additions to the build tree.
5353 * New installer built using NSIS 2.0.
5355 * Named all kernel objects in order to allow them to be monitored
5356 with tools such as SysInternals' ProcExp.exe.
5358 * Introduced new EventLog framework for AFSD
5360 * Introduced Power Management interface to AFSD for Standby and
5361 Hibernate modes to allow cache to be flushed prior to network
5364 * Utilize Win32 DNSQuery API instead of internal routines. This
5365 allows DNS SRV queries to be sent to all current domain name
5366 servers. Not just one specified in an INI file. DNS is now
5369 * "NetbiosName" registry value may be used to specify a fixed
5370 Netbios Name such as "AFS" to be used instead of "HOSTNAME-AFS"
5371 when the loopback adapter is in use. If you need to use the
5372 old notation with a loopback adapter installed specify a registry
5375 "NetbiosName" REG_EXPAND_SZ = "%COMPUTERNAME%-AFS"
5377 * Refactor all modules which depend on LAN Adapter and NetbiosName
5378 determination in a new library: lanahelper.lib. This allows for
5379 consistent behavior throughout the product.
5381 * Move the afsd.log and afsd_init.log files to the directory specified
5382 by the "TEMP" environment variable. This is usually %WINDIR%\TEMP
5383 for services. Added the Date to the log entries.
5385 * New registry value "RxMaxMTU" used to limit the size of the RX
5386 packets sent by the AFS Client Service to the Server. In order
5387 to enable OpenAFS to work across the Cisco IPSec VPN the packet
5388 size must be restricted to 1264 or smaller. The latest NSIS
5389 installer sets a value of 1260 by default.
5391 * New registry value "RxNoJumbo" to disable the use of Jumbo Rx
5392 packets. This is not needed in order to work across the Cisco
5393 VPN but might be needed for other network environments. This
5394 value is not set by the NSIS installer.
5396 * New registry value "HideDotFiles" is used to apply the Hidden
5397 attribute to files whose names begin with a '.'. This value
5398 is set by the NSIS installer.
5400 * New registry value "MaxMpxRequests" allows the maximum number
5401 of multiplexed sessions to be configured at run time. This
5402 value is not set by the NSIS installer. The default value is
5405 * New registry value "MaxVCPerServer" allows the maxmimum number
5406 of VCs per server to be configured at run time. This value is
5407 not set by the NSIS installer. The default value is 100.
5409 * New registry value "AllSubmount" allows the "all" submount to
5410 be disabled by setting its value to 0x00.
5412 * Allow cells names to be valid mount points
5413 \\<netbiosName>\<cellname>
5415 * Store the active state of drive mappings in order for afscreds.exe
5416 to restore them upon startup
5418 * Add exception handling to generate a Stack Trace to the afsd_init.log
5419 file if one happens to occur.
5421 * Add lots of logging to help detect the cause of invalid SMB packets
5423 * Enable Kerberos for Windows to be used to obtain AFS Tokens via
5424 conversion of Kerberos 5 "afs" service tickets. Supports auto-
5425 renewal of expiring tokens as long as afscreds.exe is running.
5427 * New afscreds.exe command line options:
5429 -M = renew drive maps
5430 -N = ip address change detection
5433 * New registry value "EnableKFW" in {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5434 determines whether or not MIT Kerberos for Windows should be used
5435 to obtain tokens via Kerberos 5 tickets.
5437 * New registry value "AfscredsShortcutParams" in
5438 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5439 determines the command line parameters to be specified when "fixing"
5440 the AFS Shortcut in the user's startup folder.
5442 * The "ShowTrayIcon" registry value has been moved from
5443 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5444 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5446 * The <cell name> registry values used to store the token expiration
5447 reminders have been moved from
5448 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5449 {HKCU,HKLM}SOFTWARE\OpenAFS\Client\Reminders
5451 * Obtain the Logon User Name from the Explorer key when available
5453 * new text document doc\txt\winnotes\registry.txt lists all registry
5454 values used by OpenAFS (excluding the AFS Server)
5456 * BUG: rx_securityClass objects were not properly reference
5457 counted and were never freed.
5459 * BUG: reduce the number of conditions under which CM_ERROR_TIMEOUT
5460 would be generated. The existence of a server does not imply
5461 that it is not down. If all of the servers for a cell are down
5462 return CM_ERROR_NOSUCHVOLUME instead. This prevents the Explorer
5465 * BUG: the directory name lookup cache failed to free the entries
5466 in the cache when the name cache entries cycled. The entries
5467 in the cache would become dereferenced without being freed.
5469 * BUG: fs setserverprefs could be executed without Administrator
5472 * BUG: the number of allocated NCB objects (100) exceeded the number
5473 which could actually be waited upon by the kernel (64). Any objects
5474 which were utilized above the limit could never have event completions
5477 * BUG: smb_username_t objects were not being reference counted and
5478 were not properly freed.
5480 * BUG: smb_tid_t objects could under unusual circumstances be freed
5481 before they were no longer referenced.
5483 * BUG: smb_fid_t object pointer were frequently used even when
5484 their value could be NULL. They were not properly released and
5485 therefore they were never freed.
5487 * BUG: smb_packet_t data structures were not completely initialized
5490 * BUG: when Rx produces a CM_ERROR_NOIPC error do not return "Access
5491 Denied" because that causes the Explorer Shell to try again until
5492 access is obtained. Instead return "Remote Resources" which allows
5493 the shell to move on and treat the error as transient.
5495 * BUG: when initializing the NCBreturns structure, separate Event objects
5496 were created for each NCB although a single Event object was supposed
5497 to be shared by all.
5499 * BUG: smb_dirSearch_t objects were not being properly referenced counted
5502 * BUG: smb_tran2Packet_t objects were not being properly referenced
5505 * BUG: directory path creation did not handle the case of multiple
5506 directories requiring creation in one attempt
5508 * BUG: SMB requests which required an Extended Response were ignored.
5509 This prevented some files from being written to AFS volumes.
5511 * BUG: character strings were being freed even after they were
5512 inserted into in use data structures
5514 * BUG: inconsistent usernames were used when High Security mode was
5515 enabled. (there is still much to do in this area)
5517 * BUG: pioctl() calls which require out of band RPC operations were
5518 susceptible to race conditions when performed by multiple processes
5520 * BUG: memory allocation and deallocation crossed instances of the
5521 C Runtime Library producing memory leakage and corruption in
5522 afscreds and the client configurator.