2 * Revise SMB QuerySecurityInfo Response for MS10-020
4 MS10-020 (http://support.microsoft.com/kb/980232) has caused
5 many problems for implementors of SMB 1.0 servers and applications
6 that call GetFileSecurity() without checking the return code to
7 determine if the call succeeded. The gist of the vulnerability
8 was that the SMB redirector would pass any buffer it received
9 to the application regardless of whether or not it was valid.
10 MS10-020 protects the applications by strictly validating the
11 SMB response data structure and the data in the security descriptor
14 The problem for SMB 1.0 server implementors is that there have
15 been at least three different protocol descriptions for
16 NT_TRANSACT_QUERY_SECURITY_DESC published over the last decade
17 and all of them are incomplete. Therefore, just about no one but
18 Microsoft has an SMB 1.0 server implementation that produces the
19 exact out that they are expecting to validate.
21 The end result is that in an attempt to protect applications from
22 crashing due to invalid input being passed in directly caused
23 dozens of applications to crash by not returning any security
24 descriptor data at all. Even when the applications didn't crash
25 they might not have been able to save their data. Cisco WAAS
26 and NetApp DataOnTap systems were most adversely affected and
27 they have had CIFS protocol licenses for many many years.
29 To fix OpenAFS here is what needed to be done:
31 1. Instead of returning a security descriptor that gives ownership
32 to the NUL SID, give it to the Everyone SID and set the flag
33 that states that everyone has full access.
35 2. Validate the input parameters. In particular, check to ensure
36 that the SMB file descriptor is valid and the file has not
39 3. Enforce the maximum output data and parameter counts.
41 4. Handle buffer overflow and buffertoosmall conditions
42 in the manner that Microsoft expects them to be handled.
43 In particular, note that the parameter data which is returned
44 in the SMB Data Region is not counted in the Data Count.
45 Even if MaxData is 0, we can still return parameters values
46 as long as MaxParm is large enough.
48 * Prevent use of AFSCache file contents if mapped to
51 * The Windows version of "fs newcell" did not accept any parameters
52 and behaved quite differently from the Unix version. Instead of
53 permitting new cell information to be added, the Windows version
54 simply forced the existing cell information to be reacquired.
56 This update adds a new pioctl, VIOCNEWCELL2, to support the
57 implementation of a Unix-style "fs newcell". The functionality
58 added here differs from the Unix version in the following ways:
60 1. "fs newcell" with no arguments is still accepted
61 in order to maintain compatibility with prior Windows
64 2. "fs newcell -cell <cell> -dns" instructs the cache manager
65 to add the new cell but obtain the vldb server info from
68 3. "fs newcell -cell <cell> ... -registry" instructs the cache
69 manager to add the new cell and also save the cell configuration
70 data in the registry for use the next time the service restarts.
72 4. The -vlport and -fsport options are accepted although the
73 -fsport value is currently unsupported by the cache manager.
75 * New registry value "FreelanceImportCellServDB" instructs Freelance
76 to create a mount point for every cell name listed within the
79 * Path MTU discovery for Rx is activated.
81 * Rx socket input buffer is converted to a circular buffer.
83 * Fix usage of cm_FreeServerList(). Do not set the server list
84 pointer to NULL after calling cm_FreeServerList(). Doing so
85 can result in a memory leak.
87 * Only enable Rx NAT pings on a single anonymous connection at a
90 * Fix cm_IoctlSkipQueryOptions() buffer management. Prevents a
91 potential read beyond end of memory buffer.
93 * Reduce requested privileges when reading registry CellServDB
94 to the minimum required.
96 * Add support for RPC Pipe Service NetWkstaGetInfo levels
97 101 and 102 which are called on Windows 7 and 2008-R2.
99 * Prevent integer overflow during quota percent used calculation
100 in Explorer Shell Extension (RT 126846)
102 * Generate a meaningful error if "fs listacls" or "fs setacls"
103 are executed on the Freelance root.afs volume.
105 * RXAFS_InlineBulkStat errors must be processed via cm_Analyze.
106 RXAFS_InlineBulkStatus does not return errors such as EACCES,
107 VNOVOL, VNOVNODE, VOFFLINE, VBUSY, VIO, VMOVED, etc. as an RPC return
108 code. Instead they are returned in the status info errorCode field
111 Traditionally, the error associated with the first FID in the query
112 list has been returned to the caller of cm_TryBulkStatRPC().
113 However, the error has never been processed through cm_Analyze()
114 which means that the per-vnode processing for VNOVNODE and the volume
115 global processing for VMOVED, VNOVOL, etc. has never been performed.
116 As a result, failover to other .readonly volume instances cannot occur,
117 volume moves will not be handled, and files that have been deleted
120 This patchset makes the following changes:
122 1. If an inline bulk operation has been performed and the inline
123 errorCode is a volume global error, then that error replaces
124 the RPC return code within the cm_Analyze() processing for
125 the RPC. This will affect whether or not a retry operation
128 2. The variable 'inlinebulk' is reset to 0 at the top of the
129 cm_Analyze() loop in case failover from an inlinebulk capable
130 file to an inlinebulk incapable file server takes place.
132 3. The FID that is passed into cm_Analyze() is not a real fid.
133 Instead it consists of the cell and volume but vnode = 0.
134 This ensures that the error (if any) is not applied to the
137 4. If an inline bulk operation was performed, prior to performing
138 the cm_MergeStatus() operation a vnode a check is made to
139 determine if an error was returned for that vnode. If so,
140 cm_Analyze() is called with no connection, a fake cm_req_t,
141 the fid, and the error. This permits cm_Analyze() processing
142 to be performed on the file.
144 * Show configuration pages for all types of MSI installations
146 The OpenAFS MSI installer wizard used to not show any configuration
147 pages for "Typical" and "Complete" installations. Setting the
148 workstation cell and logon options during installation required
149 selecting the "Custom" option. Many users choose the "Typical" option
150 during installation, and thus would never see the configuration pages.
151 Therefore, for these users, the workstation cell was being set to the
154 This patch makes the workstation cell and logon option configuration
155 pages visible to all types of installations (except silent
156 installations which show no UI).
158 * cm_LookupInternal creates Freelance mount points and symlinks
159 when queries cannot be found in the Freelance root.afs directory.
160 If the search name is a full cell name for which vldb information
161 can be obtained, then a mount point is added. If the search name
162 is a left-most substring or the full cell name with a dot appended
163 to it, then a symlink was created. This approach created a very
164 poluted Freelance name space.
166 This patchset makes the following changes:
168 1. Do not create symlinks with a dot appended to the cellname
170 2. Do not create symlinks where the left-most substring is not
171 a full dot separated component of the cellname.
173 3. Permit lookups to succeed when we would have created a
174 symlink in the past without creating the symlink.
176 * BPlus tree lookups are much faster than searching through
177 the native directory format on Windows because the case sensitive
178 hash tables cannot be used successfully. Permit BPlus trees
179 to be used except when called with cm_BPlusDirFoo as the action
180 function because cm_BPlusDirFoo is used to build the BPlus trees
181 from the native directory format.
183 * Symlinks are ending up in the Freelance root.afs directory that
184 end with a dot. Make sure it cannot happen.
186 * cm_FreelanceAddMount and cm_FreelanceAddSymlink is supposed to
187 return the allocated FID of the entry that was added. However,
188 cm_NameI is called to perform the lookup without forcing an update
189 of the Freelance fake directory. As a result the entry may not be
192 Force an update prior to calling cm_NameI() by using
193 cm_clearLocalMountPointChange() and cm_reInitLocalMountPoints()
196 * The Freelance fake root directory buffers were not zero-filled.
197 This results in random behavior that can cause the service to
198 terminate unexpectedly.
200 * The validation check for the response from the GetVolumeStatus
201 pioctl is incorrect. The response is not simply a VolumeStatus
202 structure but also several C strings appended to it.
204 * When flushing a file, we need to commit the file length changes
205 as well as the dirty buffers. Call cm_FSync instead of buf_CleanVnode
206 which is called by cm_FSync.
208 * Prevent rx_rpc_stats global lock from being a bottleneck in the
213 * Avoid a race when updating cell vldb server lists
214 that can result in a crash.
216 * Avoid a deadlock when managing CM_SCACHESYNC_STOREDATA
217 state operations for directory objects.
219 * Add new Windows Application Event log messages for
220 VBUSY, VRESTARTING, ALL_BUSY, ALL_OFFLINE, and ALL_DOWN.
221 Include message throttling to prevent the same message
222 from being logged repeatedly within a five second window.
224 * Reduce lock contention by waiting for cm_buf_t I/O
225 operations to complete before permitting cm_SetupStoreBIOD
226 to analyze a buffer for inclusion in a BIOD.
228 * Split the cm_buf_t flags field to separate the flags
229 that are protected by the cm_buf_t mutex from those
230 protected by the buf_globalLock. This eliminates the need
231 to hold both locks everytime the flags field is accessed.
232 Both locks were not held in the past resulting in race
233 conditions that could result in deadlocks.
235 * Add "vos setaddrs" command.
237 * Rx library lock contention avoidance between rx_NewCall and
240 * Rx library races due to inconsistent use of rx_connection
241 conn_data_lock to protect the flags field.
243 * Rx library inconsistent use of RX_CALL_TQ_WAIT which could
246 * Rx library must signal transmit queue waiters when flushing.
247 Otherwise, deadlocks can occur.
249 * In cm_UpdateVolumeLocation, avoid searching for a ".readonly"
250 volume on a numeric volume name.
252 * File buffer allocations whose offsets are beyond server EOF
253 should be locally allocated and zero filled. The file server
254 should not be issued a FetchData rpc which is guaranteed to
257 * Enable integrated logon to work with Windows 7/2008 when
258 user logons are performed with a non-Domain Kerberos principal.
260 * Add Protection Error messages to aklog output.
264 * Prevent the Explorer Shell extension from crashing if
265 symlink creation failed. (126406)
267 * A Rx level NAT ping has been implemented.
268 Add NatPingInterval registry value to
269 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
270 to permit Nat Ping to be enabled. The default value is 0 seconds.
272 * When a re-initialization is taking place, be sure to reset
273 cm_noLocalMountPoints to 0 in case someone deletes the "Freelance"
274 registry key out from underneath the service.
276 * Add krb5 error message translation to aklog, afscreds,
277 afslogon.dll, the network identity manager afs provider
280 * Mode bits aren't directly exposed by the Win32 API. We were leaving
281 them to default to 0777 when creating new files and directories.
282 This version introduces two configuration parameters;
283 'UnixModeFileDefault' and 'UnixModeDirDefault' which are DWORD
284 registry entries that are used to set the initial mode bits.
285 If the values are set to 0, then the behavior is identical to what we
288 * Minidump files are now produced with a timestamp appended
291 * An SMB request debugging monitor has been added. When activated
292 the monitor will automatically turn on trace logging if any SMB
293 request has required longer than 60 seconds to complete and will
294 then create a minidump every 60 seconds thereafter until the
299 * Restore use of DNS AFSDB and SRV records by kaserver clients.
303 * Avoid a potential Freelance deadlock during initial execution
304 of afsd_service.exe if the old ini file data has to be
307 * Three rx library corrections. (1) Idle data connection
308 processing could timeout if the send window filled and
309 took longer than the idle data timeout period for the
310 transmit window to re-open. (2) The transmit queue
311 could be emptied prematurely. A required check for the
312 queue being in use was forgotten. (3) The function that
313 is supposed to implement a wait for the transmit queue
314 to cease being busy failed to wait.
317 * Restore use of DNS AFSDB and SRV records which were
318 unintentionally disabled in 1.5.69
321 * Add a context menu to the NetIdMgr AFS Provider
324 * Prevent an empty directory Btree from being created and
325 marked as valid if cm_BPlusDirBuildTree fails.
327 * cm_BPlusEnumAlloc should not fail if the enumeration
328 contains zero entries.
330 * In cm_BPlusXXX functions, return ENOMEM if malloc() fails.
332 * Do not leak the cm_scache_t dirlock if cm_BPlusDirBuildTree
335 * fs examine should report owner and group ids as signed values.
337 * Protect buffers in smb_WriteData from simultaneous writes.
339 * Query network interface MTU values from the registry because
340 the IP Helper GetAdapterAddresses() fails to report a manually
343 * Improve performance by not dropping and reacquiring the rx_call
344 lock in rx_WriteProc* and rx_ReadProc*.
346 * When configuring the rx library, the network interface MTU
347 must be applied to both the maximum send size as well as the
348 maximum receive size as is done when rx_SetMaxMTU is called
351 There was no 1.5.67 release
354 * Convert afsauthent.dll to use the xdr_alloc and xdr_free
355 routines from afsrpc.dll in place of malloc and free.
356 This permits the two libraries to be built with different
357 C runtime libraries without crashing.
359 * Set the DOS Readonly attribute on a file/directory
360 whenever the unix mode combined with the mask 0200
361 is true. Previously there was a discrepency between
362 the mask used for testing for readonly behavior and
363 that used for setting the attribute.
365 * Disable AFSVolSync based .readonly "whole-volume callback"
366 support because the all file servers prior to 1.5.67
367 (and perhaps 1.4.12) do not properly assign a value to
368 the AFSVolSync structure in bulk status RPC responses.
370 * Improve the error output from aklog to output the value
371 from krb5 error_message() if the afs_com_err output
372 indicates an unknown value.
374 * Hold a lock on the cm_scache_t object when the smb
375 server is assigning a new clientModTime value.
377 * Permit custom build version numbers to be set
378 without modifying the src/config/NTMakefile.<platform>
381 * Add support for binary code signing with cross-signed
382 certificates and arbitrary versions of signtool.exe.
384 * Convert VBUSY and VRESTARTING to CM_ERROR_ALLBUSY and
385 do not permit them to be exposed to the smb redirector.
387 * Convert STATUS_TIMEOUT responses to STATUS_IO_TIMEOUT
388 to avoid confusion within the smb redirector.
390 * Fix the byte order assigned to port numbers associated
391 with AFSDB record lookups. They must be network byte
392 order not host byte order.
394 * Add dynamic server ranking based on RPC round trip
398 * Official support for Windows 7 and Server 2008 R2
400 * Improved Rx RTT computation following the lessons
403 * Prevent a file server bug (FetchData returning an
404 invalid length instead of zero) from causing an
405 "unexpected network error" when writing to files.
407 * Promote DNS SRV records as superior to DNS AFSDB
408 records. Support arbitrary port numbers for vldb
411 * Add "fs listacl -cmd" support.
413 * Add AFSVolSync based .readonly "whole-volume callback"
414 support. With this functionality, multiple objects from
415 a .readonly volume can have their status validated by
416 issuing a single RXAFS_FetchStatus RPC.
418 * Remove drive mapping functionality and service start/stop
421 * Remove drive mapping functionality from afs_config.exe.
423 * Adjust SMB error return codes to avoid returning
424 STATUS_TIMEOUT which results in the SMB redirector
427 * Network Identity Manager OpenAFS Provider now provides its
428 own "AFS lock" notification icon to report the status of
429 "have tokens, have no tokens, service not started, service
430 started but inaccessible". Hovering over the icon lists the
431 cells for which tokens exist (if any) and the OpenAFS version
432 number. Double-clicking executes the Network Identity Manager
435 * When merging the current status for an AFS object, ensure
436 that the volume status for that object is set to 'vl_online'.
438 * Permit cm_Analyze to request a retry even when the cm_req_t is
439 flagged CM_REQ_NORETRY if the RPC failed due to RX_CALL_DEAD.
440 The retry will force the use of a new connection.
442 * Add digital signatures to resource only dlls as required by
445 * Prevent pioctl calls from retrying indefinitely when a sharing
446 violation error occurs.
448 * Add {HKLM,HKCU}\SOFTWARE\OpenAFS\Client DWORD "ShowMountTab"
449 to restore access to drive mapping functionality in afscreds.exe
452 * Permit Rx minimum peer timeout, maximum receive window size and
453 maximum send window size to be configured via the registry.
455 * The MSI installer now sets the ARPINSTALLLOCATION as required by
458 * DNS lookups set the port numbers in network byte order within
461 * Replace use of the 16-bit compatibility API WinExec with
462 ShellExecuteEx in afs_config.exe and afscreds.exe. WinExec
463 is incompatible with Windows 7 / Server 2008 R2 UAC
467 * Short circuit background volume checks if a shutdown
468 or suspend request is received.
470 * Avoid contacting the file server in order to allocate
471 an empty buffer beyond the current length of the file.
473 * When a request for streams on a directory or mount
474 point object is received, do not offer a default stream.
475 This was affecting the saving of roaming profiles.
478 * Fix Windows 2000 compatibility. Do not call rand_s()
479 even if it is compatible with the C run time library
482 * Fix a data consistency error between the output of
483 NetWkstaGetInfo and NetServerGetInfo RPCs, specify the Lan
484 workstation group name "AFS", and report server name as
485 "AFS" instead of "\\AFS" when the caller asks for "\\AFS".
487 * Enable executables to be run from \\AFS on Windows 7.
488 Return "Name not found" instead of "File not found" when
489 a directory or file name cannot be found.
491 * Prevent cache manager from marking file server "down" when
492 the data returned in response to either RXAFS_FetchData64
493 or RXAFS_StoreData64 is invalid.
495 * Add pioctl data validation to the AFS Explorer Shell extension.
498 * After 1.5.62 afslogon.dll would no longer crash but it
499 also would not load "domain" specific configurations.
502 * An uninitialized variable was present in the symlink
503 recursion detection loop.
505 * Reverse addition of unique per file GUID during
506 Create/Open responses. Inclusion of the GUID prevents
507 Cygwin applications (and others) from accessing //afs.
509 * Treat "filename::$DATA" as a synonym for "filename".
510 This addresses the recent report of roaming profiles
511 failing to save back to the file server.
513 * When given a choice between an error from rx_Write()
514 or RXAFS_EndStoreData(), use the rx_Write() error.
515 This ensures that over quota and access denied errors
518 * If an error occurs during RXAFS_StoreData operations,
519 apply that error to all of the buffers in the BIOD.
520 This ensures that failures will not be retried if they
521 will only fail again.
523 * Do not perform a DNS lookup for a cellname if the
524 cellname prefix is _._AFS_IOCTL_._. This will avoid
525 DNS lookups when the query is for
526 _._AFS_IOCTL_._.F7E5F580200909010061TTTTNT7TT.{10E39A49-4531-4496-A08E-842D4C440D20}.
528 * Fix Freelance root.afs volume object callback processing.
529 Object synchronization and status merging bugs.
531 * Fix Explorer Shell "invalid parameter" bug introduced
534 * Log RXKAD errors to the trace log.
536 * Prevent simultaneous pioctl calls from multiple processes
537 from stepping on each other's data.
539 * Prevent simultaneous pioctl calls from crashing the
542 * Add data validation to pioctl output processing in
543 ktc_GetToken and ktc_ListTokens.
545 * Add data validation to pioctl output processing in
546 fs.exe and symlink.exe.
549 * IMPORTANT: New support for DCE RPC Services: SRVSVC and WKSSVC.
550 No longer will browsing \\AFS produce truncated share names.
552 * Registry specified server preferences did not result in
553 the CM_SERVERFLAG_PREF_SET flag being applied to the
556 * Avoid unnecessary DNS lookups of share names as cells
557 when it is known that the name cannot be a cell name.
558 Any name that does not contain a dot is skipped.
560 * When processing Dfs Referral requests, do not return an
561 error if the server is down or busy; if the volume is
562 offline; the cell vldb cannot be reached; etc.
563 These paths are still \\AFS paths and so the client
564 should not be sent to search elsewhere for resolution.
566 * CRITICAL: Prevent the cm_Daemon thread from terminating when
567 the machine enters the suspend state. IF the cm_Daemon thread
568 dies, there is nothing to execute down server checks.
570 * Prevent use of smb_StartedLock before initialization which
571 can result in a panic in the lock package.
573 * Improve error checking in afslogon.dll in order to prevent
574 NULL pointer dereferences within GetLogonDomainOptions()
575 if the Lsa operations fail.
577 * Add support to the build system to automatically update
578 a Microsoft Symbol Store.
580 * Do not access the cm_conn_t after executing the cm_Analyze
581 loops. Doing so can result in an access to memory that has
584 * CRITICAL: If a StoreData request offset is not aligned on the
585 buffer module blockSize, (offset % blockSize) bytes of dirty
586 data will not be written to the file server.
588 * CRITICAL: If a StoreData64 request is sent to a file server
589 that does not support large files (>2GB) and the client is
590 unaware that the server has no large file support, the first
591 chunksize worth of data will not be written to the file server.
592 This bug was introduced in 1.5.3. It affects all IBM AFS file
593 servers and all OpenAFS file servers older than 1.4.0.
596 * If a file server becomes inaccessible while the cache manager has
597 dirty buffers to write, the afsd_service buf_IncrSync thread can
598 attempt to use 100% of the cpu.
600 * Fix the locking that protects the cm_cell_t lists
601 (all, name, id, free).
603 * Remove cm_cell_t objects from the name and id hash tables before
604 placing them on the free list.
606 * Fix "fs newcell" which was broken in 1.5.60.
608 * Do not attempt to synchronize dirty buffers if the associated
609 volume is known to be unavailable.
611 * Log to the Windows Event Log dirty buffers that cannot be written
612 during the shutdown process.
614 * Modify behavior of a Freelance mountpoint target that does not
615 specify a cell. Instead of assuming the target volume is in the
616 Freelance.Local cell, use the workstation "Cell" specified in the
617 registry. A mountpoint target of "#root.cell." will now mean the
618 root.cell volume in the workstation cell for the current session.
619 If the workstation cell changes from "athena.mit.edu" to
620 "andrew.cmu.edu", the referenced volume will also change without
621 requiring that the mount point targets be altered.
623 * Avoid false infinite loop errors when validating the cm_cell_t
624 free list during startup.
626 * Fix two more locations where xdr_free() should be used instead of
627 free() to permit a checked built of afsd_service.exe to work with
628 release builds of afsrpc.dll.
630 * Add cm_FindServerByUuid(). Re-implement RXAFS_InitCallBackState3()
631 to permit the server Uuid to be used to lookup the server object
632 and from that determine the cell. This permits callbacks that are
633 received from alternate addresses to be processed with a known server
634 object. Previously a request from an unknown server would clear all
635 callbacks from all cells.
637 * Fix a bug that prevented optimal performance when using a non-zero
638 value for 'daemonCheckVolCBInterval'. As a reminder, when
639 "daemonCheckVolCBInterval" is set to a non-zero value, all .readonly
640 volume callbacks are automatically renewed 90 minutes before their
643 * Fix automatic ranking of vldb servers whose values are obtained from
646 * Add logging for RX CALL DEAD errors which are generated if the remote
649 * Add failover for RX CALL TIMEOUT errors when the volume is readonly
650 or the call is to a vldb server.
652 * Add registry based cell search functionality to NetIdMgr,
653 afs_config.exe, and klog.exe.
655 * afsconf_GetCellInfo() has been modified to perform gethostbyname()
656 lookups on the host names in the CellServDB instead of using the
657 specified IP addresses. This provides aklog, pts, vos, etc. the same
658 CellServDB behavior that the Windows Cache Manager uses.
660 * Completely a behind the scenes change. The snprintf() routines have
661 been completely rewritten.
663 * When updating the stat cache entry callback of a .readonly object
664 from the volume group object, update the file server reference to
665 ensure it matches the most update to date callback.
667 * Add proper support for processing callbacks from multi-homed file
668 servers. Instead of comparing servers by cm_server_t pointer,
669 compare them by UUID when the UUID is known.
671 * During a shutdown short circuit the offline volume check daemon
674 * Properly track link counts during cm_Link and cm_Unlink operations.
676 * Return the error code of RXAFS_FetchData / RXAFS_StoreData in
677 preference to an error code reported by rx_EndCall.
679 * Add "PerFileAccessCheck" registry value to permit testing against
680 experimental file servers that include per-file acl support. This
681 value is intentionally undocumented. It is not to be used by
682 production environment deployments.
684 * Log a warning to trace log if a volume group is marked as a DFS
685 File Set. The Windows OpenAFS client does not include DFS support.
687 * Fix a bug introduced in 1.5.60 that prevents the afs netidmgr
688 provider from obtaining tokens when referrals are in play.
690 * Add "fs chown" and "fs chgrp" commands to permit the owner and
691 group of objects stored in AFS to be set from Windows.
693 * Avoid performing background daemon operations when the machine is
694 going into suspend mode.
696 * Perform offline volume checks in most recently used order.
698 * Prevent crash when a data version for a cache object goes backwards.
700 * fs setquota, fs setcachesize, vos setfields, and vos create now
701 accept human readable orders of magnitude. (K, M, G)
703 * fs listquota fixed to permit large quota sizes to be displayed.
705 * rx packet resend and data packets sent counts were incorrect.
707 * Multi-thread safe library versions are now being generated and used.
708 mtafsubik.lib, mtafsutil.lib, mtafsvldb.lib, mtafsvol.lib.
710 * Microsoft SMB Redirector (mrxsmb.sys) support for ExtendedSessTimeout
711 values are now available on XP through Windows 7. Add functionality
712 to autodetect if such support is present on the machine. If so,
713 configure it if necessary and dynamically adjust the AFS Rx timeout
716 * Correct another race condition in the Rx library that could result
717 in an unexpected panic while freeing the Rx call iovq.
720 * A fix to the pioctl library to support drive substitution
721 to UNC paths. (SUBST <d:> <\\afs\cell\path>).
723 * On April 9th Microsoft released a Hot Fix for Windows Server 2003 SP2
724 that corrects a deadlock in the smb redirector and also adds new
725 functionality that permits the AFS SMB server to be given a longer
726 timeout than is normally the case. New functionality has been added
727 to configure these additional LanmanWorkstation\Parameter values.
728 (This functionality has been backported to XP SP3 and is scheduled
729 to be released on June 5th.)
731 * The BackConnectionHostNames registry value configuration was broken
732 when dynamic re-establishment of Netbios Name registrations was added.
735 * Hidden vos commands are revealed.
737 * If the "DisableLoopbackCheck" registry value is set, do not unset
738 it during the same service session.
740 * Reorganize code that prevents multiple Store operations to the same
743 * Modify IsPathInAfs test in Explorer Shell Extension and fs.exe to
744 permit broken symlinks to be treated as being in AFS.
746 * Fix vos commands that output 64-bit integer values.
748 * Cygwin Import Libraries are provided in the SDK for all OpenAFS DLLs.
749 Permits building cygwin applications against OpenAFS libraries.
751 * OpenAFS Release Notes, Administrator Guide and User Guide now installed
752 as Windows HTML Help (.CHM) files.
754 * NSIS installer does a much better job of cleaning up files left
755 over from previous installs.
757 * Fix RT#124787, a race condition between "fs flush <dir>", "fs flushvolume",
758 or "fs flushall" and on-going directory operations that can result in
759 afsd_service.exe crashing.
761 * Add support for DNS SRV records in place of AFSDB records.Â
762 _afs3-vlserver._udp.<cellname>. Priority field is used.Â
763 Weight is currently ignored.
765 * Add a method of specifying Client CellServDB information within the
766 registry that can be used to either override the CellServDB file or
767 force the use of DNS lookups for a given cell.
769 The registry schema is as follows:
771 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\
772 "LinkedCell" REG_SZ "[cellname]"
773 "Description" REG_SZ "[comment]"
774 "ForceDNS" DWORD {0,1}
776 HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\[servername]\
777 "HostName" REG_SZ "[hostname]" Default: [servername]
778 "IPv4Address" REG_SZ "[address]" Used only if gethostbyname() fails.
779 "IPv6Address" REG_SZ "[address]" <future>
780 "Comment" REG_SZ "[comment]"
781 "Rank" DWORD "0..65535" Default: 0
782 "Clone" DWORD "{0,1}" <future: server only>
783 "vlserver" DWORD "7003" <future>
784 "ptserver" DWORD ... <future>
786 ForceDNS is implied non-zero if there are no [servername]
787 keys under the [cellname] key. Otherwise, ForceDNS is zero.
788 If [servername] keys are specified and none of them evaluate
789 to a valid server configuration, the return code is success.
790 This prevents failover to the CellServDB file or DNS.
792 Only one of "HostName" or "IPv4Address" is required. "HostName"
793 is optional if [servername] is the host name.
795 * Extend registry based CellServDB functionality to
796 afsconf_GetCellInfo() interface used by aklog and many
797 other command line utilities.
799 * libafsconf.dll moved from OpenAFS\Client\Program to OpenAFS\Common
800 as it is now used by both client and server components.
803 * PriorityClass of afsd_service.exe process raised to
804 "High" to match the priority of the system services
805 that are waiting for afsd_service.exe to respond.
806 Configurable using [TransarcAFSDaemon\Parameters]
807 DWORD "PriorityClass" value. (use values from Win32
810 * Change error reporting of CM_ERROR_BADSHARENAME to
811 use the correct error (NTSTATUS_BAD_NETWORK_PATH).
812 The error we were sending would cause the smb
813 redirector to drop the connection.
815 * Change the error reporting of CM_ERROR_ALLOFFLINE
816 and CM_ERROR_ALLDOWN to return NTSTATUS_PATH_NOT_FOUND.
817 Previous we would return Bad Network Path which
818 would cause the smb redirector to cancel all file
819 handles on the same "share".
821 * A debug version of afsd_service.exe can now be used
822 with a release version of afsrpc.dll (or vice versa).
823 afsrpc.dll now implements and exports xdr_free().
825 * Removed all unprotected OutputDebugString() calls as
826 those calls can block for up to 10 seconds if there
827 is contention for the globally shared memory or if
828 the debugger fails to read data in a timely manner.
830 * Visual Studio 2008, WDK 6.1 and DDK 6.1 are now
831 supported for builds. The production releases still
835 * Discovered that XP/2003 supports a variant on recvmsg
836 called WSARecvMsg. The input parameters are mostly
837 compatible with the Posix recvmsg. This will save
838 multiple memcpy() calls per received packet.
840 * Discovered that Vista/2008 supports a variant on
841 sendmsg called WSASendMsg. The input parameters
842 are mostly compatible with the Posix sendmsg. This
843 will save multiple memcpy() calls per sent packet.
845 * Exporting rx_ReadProc32, rx_WriteProc32 from afsrpc.dll
846 and utilizing rx_ReadProc32 from within the cache manager
847 as part of the FetchData operations. rx_ReadProc32 and
848 rx_WriteProc32 are optimized versions of rx_ReadProc and
849 rx_WriteProc for use when reading 32-bit integer values.
851 * The Network Identity Manager AFS provider now reports
852 progress events. These provide a better user experience
853 with Network Identity Manager 2.0.
855 * [RT 124293] A race condition exists which permits the
856 scp field of the an smb_fid_t object to become invalid
857 after a request on the smb_fid_t has begun. Remove the
858 race by protecting all access to the scp field with the
859 smb_fid_t mx mutex and obtaining a local reference on
860 the cm_scache_t object for the length of the request.
862 * [RT 124276] If the vldb is out of sync with the contents
863 of the file servers, afsd_service will retry too many
864 times when a file server reports a volume as not being
865 present. Now if the list reported by the vldb is the
866 same as the previously seen list, then the retry is
869 * [RT 124276] Read-only volume failover was broken in
870 1.5.53 whenever accessing a volume results in VNOVOL or
873 * [RT 124276] Prior to 1.3.70 the volume server reference
874 list was not reference counted and would be prematurely
875 freed while in use. When reference counting was added
876 in 1.3.70 a bug was introduced that could result in
877 service reference list corruption.
879 * Add Windows Application Event Log warning messages for
880 "Client SMB MPX value too large" and
881 "Client SMB Buffer Size too small".
883 * Renaming of files across directory boundaries would result
884 in an invalid handle error when attempting to access the
885 files after the move.
887 * Fix the handling of Tran2 Set Path Info RPCs. Do not fail
888 when a smb file descriptor cannot be found. The whole point
889 of using a Path Info function is because an smb file
890 descriptor wasn't allocated.
892 * More edge cases in which dynamic addition of Freelance
893 root.afs entries would get the wrong FID or where the
894 root.afs directory would not be refreshed.
896 * Buffer overflow could occur if the workstation cell name was
897 longer than 64 characters. Crashes could occur in
898 afscreds.exe, afslogon.dll, and afsd_service.exe.
900 * VNOSERVICE and VOFFLINE errors were leaking and were exposed
903 * Improve consistency of the bad data version value used for
906 * Log file server uuid values as part of the cm_server object
907 when available. Dump the cm_server object list in response
910 * Optimize the performance of resetting access control lists
911 when tokens are set or removed.
913 * Fix vos.exe so that it can work with very large volume
916 * Remove symlink recursion tests and increase max symlink count
920 * Undo the "UAC manifest fix" applied to afs_config.exe.
922 * Fixed some additional race conditions in the Rx library
924 * Ensure that Freelance allocation of vnodes follow
925 the AFS convention of odd vnodes are directories
926 and everything else is an even vnode.
928 * Add Freelance logic to mount point and symlink
929 evaluation functions.
931 * Enhance smb_ParseASCIIBlock() so that it can handle
932 all of the STRING formats defined by the CIFS Technical
935 * Validate the output of smb_ParseASCIIBlock() in all
936 callers. Return CM_ERROR_BADSMB if the STRING field
937 cannot be parsed. CM_ERROR_BADSMB will cause the
938 contents of the packet to be logged.
940 * If multiple SMB Raw Write operations were taking
941 place at the same time, there could be data corruption
942 because unique event objects were not generated for
943 each Netbios receive operation.
947 * Fix a race in rxi_WriteProc which could result
948 in a freed or in use packet being passed to
949 rxi_PrepareSendPacket if the call is currently
950 in a transmit wait state. Passing a freed packet
951 will result in a panic. Passing an in use packet
952 might result in a call failure or perhaps data
955 * Kerberos referrals broke the NIM AFS provider
956 logic for automated configuration of the workstation
957 cell with a new identity. Now if a referrals
958 response is returned, assume the cell belongs to the
961 * The interface to buf_CleanAsync() and
962 buf_CleanAsyncLocked() has been changed to permit
963 errors to be propagated to the caller.
965 * During an enumeration of a directory, if the btree
966 has not yet been generated, return an error.
968 * When mapping CM_ERROR values to NTSTATUS values
969 do not treat 0 as an error.
971 * Avoid a cm_scache_t object reference count underflow
972 when renaming via the SMB interface.
974 * smb_IsMask() can be called with a non-8.3 input buffer.
975 Separate smb_IsMask() into smb_Is8Dot3Mask() for 8.3
976 input buffers and smb_IsMask() for non-8.3 input buffers.
978 * Add cm_buf_t error state to the "fs memdump" output.
980 * pioctl drive mapping checks need to take into account
981 Global Drive Mappings and SUBST assignments.
983 * Rename operations should not be permitted if the
984 destination name collides with an existing name that
985 differs only by case UNLESS the collision is the object
986 that is being renamed.
988 * Increase the maximum number of background daemon threads
989 to 64. Leave the default at 4.
991 * Set the FILE_UNICODE_ON_DISK flag when reporting the
992 file system characteristics.
994 * Report the maximum component name length as 255 even
995 for Unicode as part of the file system characteristics.
997 * Add RX debugging information to "fs memdump" output
1000 * Refcount the cm_server_t objects returned by
1003 * Fix the UAC manifest applied to afs_config.exe
1005 * rxdebug now indicates if any RX_CONN flag is set
1006 not just some of them
1008 * Change the mapping of CM_ERROR_WOULDBLOCK from
1009 NTSTATUS_LOCK_NOT_GRANTED to NTSTATUS_CANT_WAIT
1013 * Includes #GCO Public CellServDB 07 Nov 2008
1015 * Prevent separate cell entries from being created for
1016 "foobar.org" and "foobar.org." which can happen if
1017 an AFSDB record is found instead of a CellServDB
1020 * Fix a panic when checking the SMB virtual circuit list.
1021 If a VC is destroyed while the list is being checked
1022 it is possible for the next object to become invalid.
1024 * Increased logging to afsd_init.log of netbios name
1027 * Fix a DNLC panic when a UTF-16 string that cannot be
1028 converted to a utf-8 string is provided as a filename.
1030 * In the NSIS installer, delete old afslogon.dll/afscpcc.exe
1031 from %windir%\system32 and fix the WinLogon\Notify\AfsLogon
1032 location to refer directly to the new location in
1033 %programfiles%\openafs\client\program
1035 * Be more aggressive about responding to Netbios
1036 NRC_BRIDGE/NRC_NOWILD errors. Rebind the Netbios name to
1037 the LANA within the listener thread without performing a
1040 * Use the WNet Enumeration of all connected drives as a filter
1041 in the pioctl GetIoctlHandle operation. This should remove
1042 the last situations in which there was a delay when opening
1043 an Explorer Shell properties menu with the AFS Shell
1046 * Fix the parsing of CellServDB in Windows to handle linked
1047 cells as is done on Unix.
1049 * Add support for linked cells to the cache manager,
1050 netidmgr afs provider, and aklog. When two cells are
1051 linked, a volume lookup that fails in one cell is looked
1052 up in the other. This can be used for construction of
1053 test cells that override specific volumes; cell renaming;
1056 * Increase the length of cell and realm names that can
1057 be entered in the NIM AFS provider.
1059 * Actively prevent a 32-bit afsd_service.exe from
1060 executing in the WOW64 environment.
1062 * Protect against invalid IP address values in CellServDB
1067 * Fix an Rx panic caused by an attempt to double free
1068 an rx_packet when an error occurs while writing the
1071 * Fix an exception cause by a race condition when
1072 processing multiple file change notifications registered
1073 by the Explorer Shell or other applications.
1075 * Modify the Rx debug interface to use var_args and
1076 permit Rx debugging to be used on 64-bit Windows
1077 which previously had it disabled.
1081 * Automated lock hierarchy enforcement added to the
1082 osi base lock package. This permits detection of lock
1083 violations that might but often does not result in
1084 deadlocks. Enabled by default in the checked builds.
1085 May be activated in release builds via the registry
1086 value "LockOrderValidation" in the TransarcAFSDaemon
1089 When a violation is detection an assertion is thrown
1090 generating a dump file and a message indicating which
1091 locks were obtained out of order is logged to the
1092 Windows Application Event Log.
1094 * Attempt to ensure that a dump file will be generated
1095 whenever the service panics.
1097 * Removed unused lock cm_bufGetMutex
1099 * Many out of order lock acquisitions were detected
1100 using the lock order verifier and were corrected.
1102 * A race condition within the function path cm_GetSCache() ->
1103 cm_GetNewSCache() -> cm_RecycleSCache() permitted a
1104 cm_scache_t object to be issued simultaneously to two
1105 threads. This would eventually result in a panic due
1106 to the resulting undercount.
1108 * Fix interpretation of the empty string as the ioctl path
1109 to mean the current directory. "fs lsm", "symlink list",
1110 etc. now return a "not a ..." error instead of "does not
1113 * Add SMB_STRF_SRCNULTERM flag to smb_ParseStringXXX functions.
1114 to indicate that the input string is a nul terminated string.
1115 Assign ti when input strings are nul terminated instead of
1116 length counted. (from 1.5.50)
1118 * The CIFS protocol specification for handling NT_TRANSACT_CREATE
1119 does not match the observed behavor. The 'nameLength' is specified
1120 in bytes, not in characters. Fix the implementation to match
1121 the observed behavior. (1.5.50)
1123 * The cm_HaveAccessRights() attempt at deadlock avoidance by
1124 calling lock_TryRead() on the parent directory cm_scache_t rw-lock
1125 does not avoid the deadlock. Avoid the deadlock by enforcing the
1126 lock order of the lowest vnode first. Then remove the infinite
1127 loop avoidance in cm_SyncOp() that was returning an unwarranted
1128 access denied error.
1130 * File Server lock synchronization was not propertly enforced in
1131 smb_ReceiveNTCreateX() and smb_ReceiveNTTranCreate().
1133 * Another change to the short name generator. Plus and equal are
1134 not valid characters in a short name according to the
1135 RtlIsNameLegalDOS8Dot3() in kernel mode. They have been replaced
1136 with left and right brace in order to minimize the impact.
1138 * Fix a race condition that can result in a heap overwrite error if
1139 the number of known file servers or vldb servers increases during
1140 the preparation phase of a cm_CheckServers() operation.
1142 * If the SMB Authentication would fail, a LSA allocated memory
1143 buffer would be leaked.
1145 * Rewrite the cm_cell_t allocation code to permit cell vldb server
1146 lookups to fail without leaking cm_cell_t objects.
1148 * Add "fs setcrypt auth" to permit the use of rxkad authentication
1149 only mode. Not documented.
1151 * Initialize Rx mutex and lock objects once and only once as part of
1152 the DLL load process.
1154 * When reporting the Volume Name "AFS" do not include the terminating
1155 NUL in the reported length.
1157 * Refresh Volume Location information upon receipt of a VNOVOL error
1158 from the file server and avoid a race condition that prevented the
1159 updated volume location information from being destroyed immediately
1160 after it was acquired.
1162 * In response to VNOVOL and VMOVED, remove the volume id from the server's
1163 volume list so that the volume's up and down status is not affected
1164 when the machine is shutdown.
1166 * Many fixes to the Rx stack related to packet allocation and deallocation;
1167 Rx debugging; Rx packet tracking across queues; and Rx packet loss caused
1168 by overwriting the rx_call currentPacket pointer when it was already
1169 referring to a packet.
1171 * Rx Jumbograms are now off by default. To re-enable, add "RxJumbo" = "1"
1172 in the TransarcAFSDeamon/Parameters key.
1174 * "fs flush", "fs flushvolume", and "fs flushall" now destroy locally
1175 constructed directory B+ trees for flushed directory objects.
1177 * When cached pages for directory objects have been locally modified, do
1178 not permit the locally modified pages to be mixed with pages subsequently
1179 obtained from the file server. Doing so can result in directory
1182 * If Ubix errors are received in response to VL_xxx operations record
1183 the error on the server instance and permit the operation request to
1184 be retried in case another server is able to answer the query. We have
1185 seen at least one instance of a cell in which a VL server returned
1186 UNOQUORUM (possible database truncation) when the other servers were
1187 responding successfully.
1190 * Optimize the assignment of write errors to dirty
1191 buffers by applying the first error to all buffers
1192 belonging to the same object instead of trying each
1195 * Add resource version info to xstat_fs_test and xstat_cm_test
1197 * Fix a delete vs delete[] usage problem in the WiX custom
1200 * When creating a mount point, validate the volume name to
1201 make sure it exists.
1203 * When processing a mount point, pay attention to VL_BADNAME
1206 * Remove a race condition from cm_volume_t recycling that could
1207 have resulted in two threads making use of the same object at
1210 * Fix a reference count leak on smb_vc_t objects introduced in
1211 1.5.50. Permit the objects to be freed once again.
1213 * Fix the output of "fs memdump" for smb_vc_t objects that was
1216 * Add routines for debugging smb refcount errors.
1218 * When shutting down the cache manager, do not permit AFS
1219 callback requests to be processed. Doing so can result in
1220 invalid memory accesses.
1222 * Do not recycle a deleted cm_scache_t object unless the
1223 refCount is 0. Doing so can result in a deadlock and
1224 invalid memory access.
1226 * Minimize the number of RPCs sent to the file server for
1227 directory bulk status requests. Only request status for
1228 objects that we are actually reporting on and only if we
1229 do not already have status for the object and then make
1230 sure that we return as much as we can at once via the SMB
1231 interface but not so much that it exceeds the RDR timeout
1233 * Enable bos restricted operations
1235 * Fix the creation of submounts used by the afscreds and afs_config
1238 * Avoid repeated retries when attempting to obtain access rights.
1239 If the access rights were obtained successfully but do not
1240 satisfy the request, fail the request.
1242 * Fix error code reported when attempting delete a file on a
1243 readonly volume or one that is marked with the readonly DOS
1246 * Fix a short name truncation bug.
1248 * Fix a heap corruption error when reading the CellServDB file
1251 * Fix directory listings when the directory has "lookup" permission
1252 only and the directory has never been previously accessed.
1254 * Fix several tracking issues for cm_scache_t rw locks.
1256 * Ensure that all buffers used to store cell names are the
1259 * Restore Windows 2000 support (broken in 1.5.50).
1260 LOCALE_INVARIANT was introduced in XP.
1262 * Add RxUdpBufSize registry value. Default is 256K.
1264 * Avoid multiple deadlocks between cm_freelanceLock and cm_scacheLock
1266 * Do not include trailing NULs when returning file names in
1267 a directory search response.
1269 * Prevent RX from crashing the afsd_service by attempting
1270 to allocate to many buffers at once when memory is fragmented.
1272 * Avoid a deadlock when freeing cm_server_t objects
1274 * Optimize search cell file. Do not search to end of file
1275 if the cell has already been found.
1277 * Remove explorer shell extension delay on local disks.
1279 * Automatically pre-allocate 64 Rx packets per thread
1281 * Add smb lock requests and stat cache lock allocations to
1282 the output from fs memdump.
1285 * Fix Large File Support (broken in 1.5.50)
1287 * Fix "fs setacl" (broken in 1.5.50)
1289 * Fix pioctl path evaluation of drive letter
1290 mapped paths. (broken in 1.5.50)
1292 * Fix installation of VC8 Runtime library in the EXE
1293 Retail installer. (broken in 1.5.50)
1295 * Avoid a potential deadlock during Volume Location
1296 updates if all the file servers are unreachable at
1297 the time the VL update succeeds.
1299 * Ensure that rx connection object is references
1302 * Fix propagation of over quota errors during cm_FSync()
1305 Since 1.5.39 [1.5.50 released 16 July 2008]
1306 * There were no releases between 1.5.39 and 1.5.50.
1308 * FIRST RELEASE WITH UNICODE SUPPORT
1310 * Pioctl interfaces to the cache manager refactored
1311 to provide layering between the SMB specific code
1312 and the general purpose ioctl operation.
1314 * Garbage collect dead SMB virtual circuits as soon
1315 as they are no longer being referenced. This avoids
1316 problems with outstanding locks not being dropped
1317 when the virtual circuit becomes invalid.
1319 * Remove the IBM Administration Reference documentation
1320 and replace it with the OpenAFS Command Reference
1323 * Avoid calling rx_SetDeadTime and rx_SetHardDeadTime
1324 functions each time a connection is about to be used.
1325 Do not hold a lock on the rx connection object while
1326 it is being selected. This avoids a race between
1327 threads attempting to set the timeout values and
1328 removes a bottleneck that was hampering performance.
1330 * Ensure that the smb directory attribute is set
1331 for all directory objects.
1333 * Add vs2008 support to the NSIS installer scripts
1335 * Replace the VC Runtime EXE installer with the MSI
1336 installer in the NSIS installer scripts
1338 * Properly delete the VC7.1 C Runtime libraries.
1340 Since 1.5.36 [1.5.39 released 23 June 2008]
1341 * There were no 1.5.37 or 1.5.38 releases for Windows
1343 * Use Visual Studio version of vsprintf() instead of
1344 home grown version that resulted in linking conflicts
1346 * Prevent buserver, ptserver and fileserver from crashing
1347 due to non-portable strftime() format strings.
1349 * Prevent server crashes caused by rx_InitMorePackets()
1350 being called prior to allocation of all of the required
1351 synchronization objects.
1353 * Fixed a memory leak in the internal B+ tree directory.
1354 For objects with non-8.3 names, the 'longname' memory
1355 allocation would be leaked during object deletion or
1358 * Fixed a memory leak in the background daemon cm_CheckServer()
1359 routine. The memory allocated to store the server list
1362 * Obtain a missing lock around a call to cm_RemoveSCacheFromHashTable().
1364 * Correct an abstraction layer violation. cm_scache_t objects
1365 should be marked deleted in cm_Unlink() and cm_RemoveDir() and not
1366 in smb_CloseFID(). Cleanup of deleted cm_scache_t objects should be
1367 performed in cm_ReleaseSCache() when the reference count hits zero.
1369 * Prototype cm_AdjustScacheLRU() and re-implement it using osi_QAddH().
1371 * Do not reference 'smb_logp' in cm_vnodeops.c. Use 'afsd_logp'
1374 * Ensure that scache object 'nextp' pointers are set to NULL
1375 when the objects are removed from the hash table linked list.
1377 * If a dirty buffer cannot be written to the file server because
1378 of an access denied error, mark the error in the cm_buf_t
1379 structure so that it can be dealt with instead of retrying
1383 * Update CellServDB file to GCO Public 23 Apr 2008
1385 * Fix a cm_buf_t reference count leak when attempts to write
1386 dirty buffers to the file server from within cm_IncrSyncer()
1389 * Prevent udebug from crashing.
1391 * Activate RX Idle Timeouts. If the file server is busy for more
1392 then 30 seconds, attempt to failover to another server without
1393 marking the busy server down.
1395 * Another VNOVNODE issue fixed. When writing a dirty buffer
1396 to the file server, if VNOVNODE is received, mark all buffers
1397 as invalid without further attempts to contact the file server.
1399 * Improved performance on high latency links.
1402 * Fix a bug preventing the re-initialization of the Freelance
1403 root.cell contents when a change is made.
1405 * Properly return CM_ERROR_NOSUCHFILE when removing a symlink
1406 or mountpoint that does not exist.
1408 * aklog, afscreds, integrated logon, and the NIM AFS credential
1409 provider will all attempt to first acquire an afs service ticket
1410 using the service principal: afs/<cell>@<CLIENT_REALM>. This
1411 permits the proper detection of the cell realm when multiple
1412 realms are local realms for the cell.
1414 * Fix fs commands that re-use file identifiers instead of forcing
1415 a path evaluation each time. This was broken in 1.5.33 when
1416 hashes were added to the cm_fid_t structure.
1418 * Fix symlink and mountpoint removal and then recreation when
1419 applied to the fake Freelance root.afs volume.
1421 * Fix SMB change notifications for the Freelance root.afs volume
1422 when creating or removing symlinks and mount points.
1424 * Force DNS AFSDB cell name lookups to lower case if the DNS
1425 response is not all lower case.
1427 * 32-bit installers are now built with Visual Studio 2005
1429 * All vos.exe commands now support the "-noresolve" option which
1430 forces the display of IP addresses instead of DNS hostnames.
1432 * No longer hold the cm_volume_t mutex across VL_xxx RPCs. This
1433 permits operations that do not require up to date volume
1434 location data to continue while waiting for the RPC to complete.
1435 Multiple threads waiting for a volume location update on the
1436 same volume set no longer trigger multiple calls VL_xxx RPCs.
1438 * Volume server references could be marked offline and never
1439 reset to the not_busy state.
1441 * Further optimizations to cm_Analyze() retry logic for
1442 CM_ERROR_OFFLINE and CM_ERROR_ALLBUSY error handling.
1445 * Optimize the DNLC by applying Interlocked incrementing for
1446 statistics gathering and enabling greater use of read
1447 locks instead of write locks
1449 * Further optimize osi_Log operations when logging is disabled
1450 by removing the need for the function call overhead.
1452 * Add a new lock operation, Convert Read to Write, and apply
1453 it throughout the cache manager. This function provides a
1454 fast transition from a read lock to a write lock when the
1455 caller is the only reader.
1457 * Fix a timeout problem when opening files or request locks
1458 when the file in question is already write-locked by another
1461 * Fix a deadlock that can occur if "fs flushall" is executed
1462 while applications are writing to AFS.
1464 * The cache manager would re-read data from the file server
1465 after file truncation when obtaining buffers to write to the
1466 same file until the first StoreData RPC completes.
1468 * Fixes error handling in cm_NTCheckDelete()
1470 * Replaces cm_scache_t mutex with a read-write lock permitting
1471 additional parallel access.
1473 * Prevent an cm_scache_t reference undercount when evaluating
1474 symlinks containing @sys when none of the active sysnames
1475 match any of the available targets.
1477 * Prevent the leak of a cm_scache_t rwlock if associated symlink
1480 * Set RxMaxMTU registry value to 0 which is the equivalent of
1481 it not being set at all.
1483 * Reduce contention between BkgDaemon threads
1485 * Prevent CM_SCACHEFLAG_ASYNCSTORE from being reset on a write
1488 * Convert reference counts on cm_volume_t, cm_conn_t, cm_server_t
1489 objects to use the faster Interlocked operations which avoid
1490 the need to hold write locks.
1492 * Permit rxkad errors other than RXKADEXPIRED to be treated as a
1493 non-fatal error that forces a retry against another service
1496 * Fix a crash in smb_WriteData that would occur if an application
1497 opened the file as read-only and then sent a write request anyway.
1500 * Respond to VL_NOENT errors by removing cm_volume_t object and
1501 setting it to be the first object to be recycled. This is not
1502 only a more efficient use of resources but it also prevents
1503 repeated attempts to query the VLDB server for the volume from
1504 the CheckOfflineVolumes routine.
1506 * Change the defaults for RxEnablePeerStats and RxEnableProcessStats
1509 * A first cut at a cache manager statistics monitor that can be used
1510 to determine the necessary cache parameters to support the
1511 application working set.
1513 Off by default, the performance package can be activated by setting
1514 "daemonPerformanceTuningInterval" in the service Parameters key.
1515 As with the other daemon interval values the unit is in seconds.
1517 At service start and each succeeding interval the cache manager
1518 will write statistics to %TEMP%\afsd_performance.txt showing the
1519 relative usage of cm_scache_t, cm_volume_t and cm_buf_t objects.
1520 The FID statistics keep track of all FIDs seen by the cache manager
1521 during the service session whether or not they are backed by any
1522 live object in the cache.
1524 These statistics are not stored in the cache file.
1526 * Fix windows event logging broken in 1.5.33.
1528 * Eliminate multiple calls to syscfg_GetIFInfo() which is used to
1529 populate the IP address list for WhoAreYou/TellMeAboutYourself RPCs
1530 as well as when computing the server preference list.
1532 * When merging status after a StoreData operation prevent data
1533 buffers from being discarded that we would prefer to keep.
1534 (Bug introduced in 1.5.33.)
1536 * Modify installers the set RxMaxMTU to 0.
1538 * In the AFS Configuration Panel on the Advanced->Miscellaneous
1539 dialog, the Background Daemons and Service Threads fields were
1542 * Increase default number of Background Daemons to 4.
1544 * Remove cm_volume_t reference from the cm_scache_t object. Permit
1545 cm_volume_t objects to be recycled.
1547 * Windows Error Reporting has shown a number of previously unseen
1548 stack corruption errors in the Explorer Shell extension. The
1549 minidump appeared to indicate the problem was in ParseAcl().
1550 Reviewing the function it appears that numerous things could go
1551 wrong if invalid input was provided. Added error checking and
1552 data validation. Hopefully this will address the problem.
1554 * Windows Error Reporting has shown crashes in aklog as a result
1555 of krb5_cc_get_default() failures not being handled. They are
1559 * The Rx library used a 32-bit type for sockets which was
1560 truncating the socket value on 64-bit Windows. This
1561 was not an issue in general but became an issue when
1562 Rx "hot thread" support was enabled.
1564 * Enable "Rx hot thread" support which ensures that while
1565 a received message is being processed that a new thread
1566 is allocated to receive the next message to arrive.
1568 * Increased parallel access to smb_rctLock protected
1569 resources. Reduce the number of times that the
1570 lock is dropped and re-obtained.
1572 * Increased parallel access to cm_scache_t and cm_buf_t
1573 resources by replacing the previous global write lock
1574 protected reference counting mechanism with one built
1575 upon the use of InterlockedIncrement and InterlockedDecrement.
1577 * Implement a hash for cm_fid_t to reduce comparison costs
1578 and inline the cm_FidCmp() function.
1580 * Redefine the BUF_HASH and BUF_FILEHASH in terms of the cm_fid_t
1581 hash which has a better distribution
1583 * Modify cm_ConsiderPrefetch to evaluate the amount of data in
1584 the most recent read request instead of the next chunkSize.
1585 cm_chunkSize can be dozens or hundreds of buffers. As a result
1586 too much time is spent performing the evaluation.
1588 * Fix the usage of cm_scache_t bufCreateLock. The purpose of this
1589 lock is to prevent the creation of new buffers while a truncation
1590 is being performed. All references to bufCreateLock have been
1591 removed except in two places:
1592 i. a write-lock surrounding the function that calls buf_Truncate()
1593 ii. a read-lock within buf_GetNewLocked() that actually allocates
1596 * Modify the CM_CONFIG_DATA_MAGIC value to include a
1597 CM_CONFIG_DATA_VERSION number which value be used to force the
1598 replacement of the cache file contents when incompatible changes
1599 are made between releases.
1601 * CM_SCACHESYNC_ASYNCSTORE should not be ordered by
1602 cm_SyncOpCheckContinue
1604 * Avoid calls to multi_Rx if nconns == 0
1606 * Modify smb_WriteData to perform background writes based upon the
1607 crossing of 'smb_AsyncStoreSize' boundaries instead of cm_chunkSize
1608 boundaries. This will slow down writes from the SMB interface but
1609 will avoid the risk of the CIFS client disconnecting from the AFS
1610 client SMB server. The default value for 'smb_AsyncStoreSize'
1613 * SMB AsyncStore functionality is on by default but can be
1614 disabled using the registry value:
1616 HKLM\SOFTWARE\OpenAFS\Client DWORD "EnableSMBAsyncStore"
1618 smb_AsyncStoreSize can be adjusted with
1620 HKLM\SOFTWARE\OpenAFS\Client DWORD "SMBAsyncStoreSize"
1622 The value must be a multiple of the buffer block size
1623 and cannot be larger than the chunk size.
1625 * Do not leak a read lock on the bufCreateLock
1627 * Do not forget to call cm_SyncOpDone if cm_SyncOp succeeded
1630 * In MergeStatus, only remove clean buffers from the hash table.
1632 * Avoid the need to update the data version number on each buffer
1633 associated with a scache when MergeStatus is called after a
1634 StoreData by maintaining a range of valid data versions as part
1635 of the cm_scache_t object.
1637 * Profiling shows large numbers of blocked calls in cm_HoldSCache
1638 from cm_HaveAccess... We can safely avoid the call so do so.
1640 * Add "ForceAnonVLDB" service parameters registry value. Default
1641 is 0. Set to 1 to force the use of anonymous connections to
1642 the VLDB servers. This is useful when using krb5 tokens and the
1643 VLDB servers have not been updated.
1646 * Do not attempt to search for a cell name that matches one
1647 of the special share names "svrsvc", "wkssvc", "ipc$"
1649 * Issue permission denied warnings from afscreds if attempts
1650 to modify the startup shortcut or registry values fail.
1651 This happens under Vista UAC.
1653 * Do not return BAD_NETWORK_ERROR in response to attempts
1654 to create directories or files in the Freelance root.afs
1655 volume. Instead return ACCESS_DENIED.
1657 BAD_NETWORK_ERROR was being sent because the Freelance cell
1658 has no vldb servers. Short circuit the error by testing for
1659 freelance in CreateFile and MakeDir
1661 * Treat srvsvc, wkssvc and ipc$ as pseudo files. Permit them
1662 to be opened and closed without actually existing. For now
1663 we treat them as equivalent to ioctls but this should
1666 * Fix RAP processing to not reject names that exist in the
1667 root.afs volume for GetInfo queries.
1669 * In afscreds, use GetUserNameEx to obtain the username instead
1670 of searching the registry. Try UserPrincipalName first and
1671 fallback to SamCompatible.
1673 * Stop linking to rpcns4.dll which is no longer supported on
1674 Vista and 2008. TaAfsAdmSvr is going to have to be
1677 * When processing dfs referral requests, if the sharename is
1678 a partial match and therefore will not be added automatically
1679 to the share list, make sure that we return an error.
1681 * Avoid a crash when logging VNOVOL errors returned from
1682 cm_CheckOfflineVolume.
1684 * When trace logging is disabled, do not permit osi_LogSaveString
1685 to perform the string copying. By disabling the string
1686 copies, throughput of the service more than doubles.
1688 * when checking offline volumes, update the cell vlserver info
1689 if required and don't perform a check if there are no servers
1692 * Add volumes in the vl_unknown state to the list of objects to be
1693 checked as offline or retried when an allbusy error is received
1696 * Add support for configuring rx_ExtraPackets
1698 * All calls to buf_Get() must be protected by a read lock on the
1699 cm_scache_t bufCreateLock
1701 * When a MergeStatus results in the invalidation of buffers, do
1702 not remove buffers from the hash tables that have active references.
1704 * Upon receiving a VNOVNODE error from a file server, be sure to
1705 invalidate any file locks. This prevents on-going attempts to
1706 extend the locks and will cause subsequent attempts to access
1707 the file to receive a invalid handle error.
1709 * Prevent smb_ReceiveTran2SearchDir from adding terminating NULs
1710 to directory entry names.
1712 * No longer discard tokens when RXKADBADTICKET error is received.
1713 Instead return STATUS_NO_KERB_KEY error.
1716 * Add "cmdebug <host> -cellservdb" option and client service
1717 implementation permits querying the vldb server for each
1718 cell known to the cache manager. The output is in a form
1719 that permits it to be used as a CellServDB file.
1721 * Add Symlinks->Show option to the AFS Explorer Shell Context
1724 * Prevent buffer overrun in the cmd parser used by command
1725 line tools such as fs, vos, pts, etc.
1727 * The path ioctl operations have several issues:
1729 (1) the specified path for "fs examine, whereis, whichcell,
1730 flush" and so always has the follow symlinks and mount
1731 points semantics. This makes it impossible to determine
1732 what the FID of a symlink or mount point is.
1734 (2) "fs examine" out is not the result of a single pioctl
1735 operation but is actually the combined output of half
1736 a dozen operations. Path evaluation is an expensive
1737 operation. It would be faster if the caller could
1738 evaluate the FID first and then perform all of the
1739 rest of the operations by specifying the FID instead
1742 (3) fs output reports all objects as files. By adding a
1743 GetFileType pioctl more informative output can be provided
1744 that indicates what type of object the path evaluates to.
1746 (4) the Windows fs command includes a number of commands that
1747 do nothing but exist only because the Unix cache manager
1750 A new extendible data structure cm_ioctl_query_opts_t has been
1751 added which can be optionally specified with pioctls that do not
1752 already require input data. The first two fields of this structure
1753 are 'literal' and 'fid'. The literal field is used to indicate
1754 whether the last component of the path should be evaluated following
1755 symlinks and mount points.
1757 The fid field permits a fid to be specified.
1759 * A new GetFileType pioctl has been added. The type of objects are
1762 * A new "-literal" option is available for "fs examine, flush, whereis,
1765 * Unimplemented fs commands have been removed.
1767 * There are circumstances where a volume object is being accessed and
1768 the volume is marked indicating that the volume location information
1769 is out of date but where it is also pointless and perhaps dangerous
1770 to block waiting for the rpc to complete. One example is when
1771 processing the cmdebug requests. If we know that we are not going
1772 to use the volume object to contact a server, then we can now set
1773 the CM_GETVOL_FLAG_NO_RESET flag.
1775 * when moving up in the directory tree we search the recorded fid list
1776 to find a matching fid that we have already crossed. we must also
1777 reset the fid count based upon what we discover.
1779 * Cell names are published as share names. As a result they are searched
1780 for as part of the Dfs Referral evaluation. If share "foo" can not be
1781 reached, the CIFS client will fallback to searching for "fo" as well.
1783 Since the freelance client automatically adds a symlink for prefixes
1784 of the cell name, this results in both "foo" and "fo" being added to
1785 the freelance root.afs volume. This patch prevents that negative side
1788 * when tracking Kerberos credential cache names be sure to include
1789 the cache type prefix.
1791 * Microsoft has assigned OpenAFS a network type value
1793 * The algorithms that were used to produce a Netbios name were broken.
1794 The name that was produced when the hostname was too long was in fact
1795 longer than the maximum permitted netbios name.
1797 Also, the "NetbiosName" value was not used as a suffix when the loopback
1798 adapter was not installed. The hardcoded string "AFS" was used instead.
1800 * avoid another deadlock during server probes initiated by the ipaddr
1801 change daemon thread.
1803 * prevent the afs client service from crashing when fs newcell is
1806 * In the NSIS installer, move afslogon.dll, afscpcc.exe and afs_cpa.cpl
1807 to \Program Files\OpenAFS\Client\Program in order to get them out
1808 of the %windir% tree
1810 * EDQUOT == WSAEDQUOT. Define it in the right places so that in the
1811 end STATUS_OUT_OF_QUOTA can be returned to the application.
1813 * Modify the search order for determining the location of CellServDB
1814 and other client configuration files.
1816 1. AFSCONF environment variable
1818 3. NEW - use All Users\AppData\OpenAFS\Client only if CellServDB exists
1819 4. use Program Files\OpenAFS\Client
1821 * More Vista Power Management tweaks
1823 * Close a socket handle leak in the kauth routines
1825 * Don't hold a cm_volume_t mutex across RPCs
1827 * Convert downTimes to use clock time instead of relative times.
1828 This permits correct comparisons with expiration times.
1830 * Update the Wix installer to properly install the prerequisites
1831 for translate_et.exe
1833 * Do not translate pioctl paths as they are not converted by the
1836 * Implement multi_rx probes for checking whether servers are up
1837 or down. Turns probes into a constant time activity instead
1838 of one dependent upon the number of known servers.
1840 * Treat VNOVOL errors the same as VOFFLINE. Do not force an
1841 update of the volume location data and retry. VNOVOL means
1842 that the volume could not be attached. It doesn't mean that
1843 the volume is not on this server.
1845 * Modify the Explorer Shell AFS Context menu Symlink->Add to
1846 not attempt to validate the target path. There is no reliable
1849 * Improve performance of \\afs\<cell><mount-type><volume>\
1850 access by eliminating extraneous attempts to search the
1851 CellServDB file and DNS AFSDB records. Share names
1852 containing '%' or '#' are not valid cell names.
1854 * Return STATUS_RANGE_NOT_LOCKED if the application attempts
1855 to unlock a range that is not currently locked.
1857 * Enable the 32-bit tools to read the 64-bit registry keys
1858 when executing in the WOW64 environment.
1862 * Fix buffer refcount leak introduced in 1.5.29
1864 * Prevent the NIM AFS Provider from crashing when configured to use
1865 Kerberos v4 for token acquisition in situations where no Kerberos v4
1866 support is available. For example, 64-bit KFW or sites that delete
1867 the Kerberos v4 DLL.
1869 * In afsd_init.log, support logging cache sizes greater than 2GB.
1871 * The volume status output of "fs examine" was being determined based
1872 upon the wrong error value. Instead of using the pioctl() return
1873 code, the errno value must be used.
1875 * Prevent invalid pts auto-registration attempts by aklog and the
1876 NIM afs provider when Kerberos referrals are in use. (krb5 1.6+)
1878 * Add synchronization protection to all of the SMB Listener State
1879 variables. This prevents race conditions when the addition or
1880 loss of a network adapter takes place.
1882 * On Vista, there is a race condition between the restart of the
1883 afs client service and the network adapters. Ensure that the
1884 afs client service can handle switching between loopback and
1885 non-loopback modes of operation.
1887 * Fix FollowBackupPath to work with mount points that are not
1888 explicitly read/write.
1890 * Fix a deadlock that has been observed on Vista while resuming
1891 after at least four hours of sleep.
1893 * Add new HKLM registry configuration options that can be used
1894 to provide configuration hints to the NIM AFS Provider when
1895 creating new identities. See release notes for details on
1896 HKLM\SOFTWARE\OpenAFS\Client\Realms\.
1899 * Speed up write buffer operations by holding a global read-lock
1900 instead of a write-lock when updating the buffer data version
1901 number. The data version is protect by the associated stat
1904 * Speed up read buffer operations by removing out of date buffers
1905 from the buffer hash tables.
1907 * Protect against a null pointer dereference in afslogon.dll
1908 if QueryAdHomePathFromSid() fails.
1910 * Remove a race condition while accessing private freelance data
1913 * Update the buffer data versions when Freelance directory data
1916 * Complete the implementation of smb_ReceiveTran2GetDFSReferral().
1919 * Return EAS_NOT_SUPPORTED in response to EA read/store/query
1920 instead of returning zero length.
1922 * Return ERROR_RETRY when a server sends VBUSY.
1924 * Return PATH_NOT_COVERED instead of BADSHARENAME for DfsLinks when
1925 a volstat plug-in is in use.
1927 * Enhance the volstat interface to support Dfs path mappings.
1929 * Fix "symlink list" to do the right thing for Dfs links
1931 * Fix callback expiration that was broken in 1.5.27.
1933 * Fix a crash at shutdown caused by the lack of synchronization
1934 within the daemon threads (cm_Daemon, cm_BkgDaemon, cm_IPAddrDaemon)
1936 * Hold the cm_server_t mutex while walking or modifying the
1937 cm_server_vols_t list in order to avoid a race condition.
1939 * Ensure that the cm_server_t refCount is safely modified.
1941 * The pthread library implementations of pthread_cond_wait
1942 and pthread_cond_timedwait were failing to return with
1943 the mutex held when error conditions occurred. This could
1944 result in race conditions wherever they are used. One case
1945 in which problems were seen was the Rx event_handler().
1947 * Prevent cm_GetBuffer() from locking all buffers associated with
1948 a stat cache object before deciding if the current buffer is up
1951 * Fix cm_BkgPrefetch to actually ensure that there is a cm_buf_t
1952 object for each offset that we will attempt to reserve.
1954 * Reduce the default number of background daemon threads used to
1955 prefetch or store to one in order to prevent two or more threads
1956 from stepping on each other.
1958 * Reduce the default chunk size to 256KB in order to prevent read
1959 timeouts on low bandwidth connections.
1961 * Fix a race condition that could result in threads becoming stranded
1962 when synchronizing multiple requests on the same stat cache object.
1965 * The GiveUpAllCallBacks functionality that was added in 1.5.21
1966 has been disabled. It has been disabled because it turns out
1967 that the RPC is improperly implemented in all file servers
1968 previous to 1.4.6. The RPC is executed without the proper
1969 locks being held resulting in data corruption and eventual file
1970 server failure. The functionality will be re-enabled in a
1973 * a couple of .readonly Volume Callback optimizations were
1975 - Apply most recent volume callback to all stat cache objects
1976 in the volume. This avoids premature callback expirations
1977 and unnecessary FetchStatus calls.
1978 - Add option to permit attempts to renew .readonly Volume
1979 Callbacks on a periodic basis to avoid invalidation of stat
1980 cache data. [Requires daemonCheckVolCBInterval be set to
1981 non-zero number of seconds. 1800 (30 minutes) is suggested.]
1983 * Adds support for 64-bit data version values. Previous releases
1984 only supported 32-bit values. Larger values sent by the file server
1985 would result in rollover.
1987 * aklog and asetkey no longer generate an exception if Kerberos for
1988 Windows is not installed.
1990 * If the LSA authentication functions report a temporary out of memory
1991 condition, fail the current authentication and do not cause the AFS
1992 client service to terminate unexpectedly.
1994 * When the AFS client service is shutting down, prevent the daemon check
1995 thread from executing additional checks which might attempt to access
1996 discarded locks or unmapped memory.
1998 * Correct a double free condition in the NIM AFS credentials provider.
2000 * Add "FollowBackupPath" service parameter registry value.
2001 This provides equivalent functionality to the UNIX afsd -backuptree
2002 option. When set to a non-zero value, normal mount points originating
2003 in a .backup volume will prefer a .backup volume.
2006 * Fix NSIS uninstall from the .MSI installer.
2008 * Add Stack Frame logging for AMD64
2010 * Add support for EWOULDBLOCK to cm_Analyze. If the file
2011 server returns EWOULDBLOCK, retry the request every two
2012 seconds for up to the RDRtimeout.
2014 * Reorganize the order of the includes to ensure that EWOULDBLOCK
2015 is not assigned the same value as EIO.
2017 * Update CellServDB to GCO Public 25 Oct 2007
2019 * Add additional validation and error handling code after
2020 each call to getSlot(). If an invalid slot is returned,
2021 return NONODE. If the invalid slot is returned when
2022 extracting a data node, invalidate the tree.
2024 * Modify compareKeys() to always perform a case-insensitive
2025 comparison and only perform a case sensistive comparison
2026 if the case-insensitive one matches. This ensures the
2027 ordering is consistently reported.
2029 * Add lock assertions to ensure that all calls are being
2030 performed with the correct locks being held. There have been
2031 some crash reports that provide stack data that does not appear
2032 to be possible unless there is a race. However, there are no
2033 obvious locations where the race is taking place and the test
2034 suite indicates that all of the correct locks are being held.
2035 We shall see what happens in the field.
2037 * For consistency replace all calls to findKey in which the range i
2038 s (1,numentries) with calls to getSlot().
2040 * Optimize the depth search loop by testing the slot value in the
2041 for statement instead of forcing the loop to be broken later.
2043 * Reorganize the locking for cm_BeginDirOp and cm_EndDirOp.
2044 There are a number of locations where locks are obtained, dropped,
2045 and reobtained. This reorganization attempts to accomplish several
2048 (1) be optimistic for the most common case so it will be fast
2050 (2) add consistency checks after each location where locks are
2051 dropped and re-obtained. If we lose a race in cm_BeginDirOp
2052 and the bplus tree is out of date, retry until we get to a
2053 consistent state that we can use.
2055 (3) Ensure that all operations take place with the correct locks.
2057 * One of the issues that has become a serious problem since the
2058 addition of the local directory updates is that although cm_SyncOp
2059 synchronizes operations, it does not preserve the order of requests.
2060 This has always been a problem in that it has been possible for a
2061 request to fail to complete due to its worker thread's bad luck.
2062 When a request takes longer than the Windows SMB Redirector's timeout,
2063 the SMB Redirector tears down the SMB virtual circuit.
2065 When using the local directory updates it is really important that
2066 the directory update operations complete in the order that they were
2067 sent to the file server. If they don't, then the local directory state
2068 and the file server state will not match and the local directory state
2069 must be discarded which in turn forces a new read of the entire
2070 directory contents over the network.
2072 This patch adds a new cm_scache_waiter_t object that is used to store
2073 the current thread, buffer, and syncop flags within a waiters queue
2074 on each cm_scache_t object. If a thread is forced to sleep in
2075 cm_SyncOp, upon waking it will check to see if there are any other
2076 threads waiting that are attempting to perform a similar task ahead
2077 of it in the queue. If yes, the thread goes back to sleep. If not,
2078 it goes ahead and enters the cm_SyncOp conflict resolution block.
2080 This patch has the additional side effect of reducing the number of
2081 competing threads that must obtain the cm_scache_t mutex and process
2082 the cm_SyncOp conflict resolution block. As a result, the overall CPU
2083 utilization of the service and the clock time associated with processing
2084 requests will be reduced.
2086 * assert that the cm_scache_t mutex is held when calling
2087 buf_ForceDataVersion().
2089 * rename findNode to leafNode in bplus_Lookup
2091 * replace all OutputDebugString calls with osi_LogX calls
2093 * modify bestMatch to special case the return values for leaf nodes.
2094 If an entry is above or below the values available in the leaf node
2095 return BTLOWER or BTUPPER instead of BTERROR.
2097 * In insert and delete operations check for BTLOWER/BTUPPER and isleaf,
2098 if true convert to either slot 0 or Max and perform the insertion.
2099 This produces easier to read code when performing lookups.
2101 * modify lock_AssertXXX macros to call osi_assertx() and provide a
2102 descriptive message.
2104 * the difference between osi_assert() and osi_assertx() is whether or not
2105 a description is specified as part of the call. When no message is
2106 specified afsd_notifier() is called with a NULL msgp parameter. This
2107 results in a null pointer dereference during the EventLog() call.
2109 As a result, none of the information describing the assertion is written
2112 This commit sets a default message for use when no other message is
2113 provided and it replaces all calls to osi_assert with osi_assertx and
2116 * Migrate B+ tree search key into thread local storage
2119 * Fix an uninitialized variable in lana_IsLoopback() which
2120 could result in a random determination that a loopback
2121 adapter is not a loopback adapter. This would result in
2122 a computed Netbios service name of %MACHINE%-AFS instead
2123 of AFS. This is a serious problem if the value computed
2124 by the AFS Client Service and the client process pioctl
2125 calls produce different results as it will result in a
2126 communication failure.
2128 * Fix a crash in the B+ Tree code that results when a file
2129 rename only results in a change of case.
2131 * Fix the pioctl path parsing. The previous algorithm
2132 broke the current directory of the client into two parts,
2133 the TIDpath which represented the device root and the
2134 relative path. The problem is that this is not sufficient
2135 from proper processing of symlinks and mount points.
2137 What is required is that the last component of the path
2138 must not have the symlink or mount point be evaluated
2139 whereas symlink and mount point evaluation must be performed
2140 for the rest of the relative path. Therefore, the path is
2141 now split into three parts: TIDpath, intermediate path,
2144 "fs examine" now reports the data of the symlink or mount
2145 point object instead of the target and "symlink list" and
2146 "symlink remove" now works. The "Explorer Shell Extension"
2147 menu now is displayed for symlink objects.
2149 * More informative symlink.exe error messages.
2151 * When the cmd.exe "del" command is used to remove a symlink,
2152 not only was the link object removed, but the stat cache
2153 object for the target was marked deleted. This resulted
2154 in cache corruption which would require the destruction
2155 of the cache to correct.
2157 * Add undocumented registry value, "OfflineReadOnlyIsValid"
2158 which results in cached contents from read only volumes
2159 to be treated as valid when there is no callback registered
2160 and all servers are down or all instances are offline.
2162 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
2164 * Add documented registry value, "AcceptDottedPrincipalNames"
2165 which is used to permit the acquisition of AFS tokens
2166 using Kerberos v5 principal names which include a dot in
2167 the first component.
2169 {HKLM\HKCU}\Software\OpenAFS\Client
2171 * "fs uuid" does not require "AFS Client Administrator" group
2172 membership to display the current UUID value.
2175 * Update NSIS installer to version 2.30
2177 * Update Wix installer to version 2.0.5325
2179 * 64-bit KFW support:
2180 - Network Identity Manager plug-in
2182 - 64-bit krb5 enable afscreds.exe, afs server manager,
2185 * Enable B+ tree Directory Search by default.
2187 * NIM plug-in is now compatible with FILE: ccaches
2189 * Install instloop.exe into Client/Programs/ even though
2190 it isn't required. Permit it to be used by organizations
2191 that clone Windows systems.
2193 * On a suspend, only drop callbacks for servers that
2194 are online. If the server is offline, maintain the callback
2195 until the server is once again available.
2197 * Prevent against acquisition of a directory write lock when
2198 there are no updates to be performed. This was resulting
2199 in a write-lock leak which later on would lead to a deadlock.
2201 * Add configurable file pre-fetch mode. For files that match
2202 the specified file extensions, pre-fetch the entire file into
2205 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2206 (MULTI_SZ) PrefetchExecutableExtensions
2208 * Fix the client_osi large integer support so that return values
2209 can be properly assigned to large integer variables.
2211 * Do not hold cm_scache_t mutex across cm_GetCell() calls
2212 since cm_GetCell() can block in a RPC.
2214 * Add short file name support to B+ tree searches.
2216 * Resolve a conflict between the desire to notify file servers
2217 that the client workstation is suspending and the desire to
2218 permit objects with valid callbacks when the client suspends
2219 to continue to use those objects from the cache if the client
2220 resumes without network connectivity.
2222 * Fix stack variables to ensure that all buffers storing paths
2223 can handle AFSPATHMAX (1024) characters. Paths of 256 characters
2224 would cause a stack overflow condition.
2226 * On Vista, force afs_config.exe to run with the highest privileges
2227 available to the end user. For a user in the Administrators
2228 group running under UAC, this will cause afs_config.exe to be
2229 be run with full Administrators privileges after the user
2230 approves the privilege escalation.
2232 * More VNOVNODE handling issues. If a dirty buffer cannot be
2233 written due to VNOVNODE, clear the buffer and ignore the error.
2234 If a cm_scache_t object has been marked deleted due to a VNOVNODE
2235 error, the next time the smb_fid_t object that references it is
2236 used return either NOSUCHFILE or NOSUCHPATH and close the
2237 smb_fid_t object. Subsequent attempts to use the smb_fid_t will
2238 result in an invalid handle error.
2240 * For Vista, set the RequestExecutionLevel of the NSIS installer
2243 * For the build environment, support the generation of manifests
2244 when using the CL1310 (VS .NET 2003) compiler/linker.
2246 * There was a bug in OutputDebugHexString() which could overflow
2247 a stack buffer resulting in unexpected process termination
2248 of afsd_service.exe if the input Security Token contained
2249 many bytes equivalent to the '%' char.
2251 * Add directory lookup statistics to "fs memdump" output.
2255 * Windows uses case-insensitive file name pattern matching
2256 but AFS is a case sensitive file system. The AFS3 directory
2257 format is block based, uses network byte order and
2258 includes a hash table for fast case sensitive lookups.
2259 This causes several problems for the Windows AFS client.
2260 (1) Traversing the directory blocks is cpu expensive
2261 (2) A hash table miss does not indicate that the desired
2262 entry does not exist.
2263 (3) Determining whether a non-ambiguous inexact match or
2264 the entry does not exist requires a linear traversal
2265 of the entire directory.
2266 These issues often result in 100% CPU utilization.
2268 These issues are addressed by building a modified B+ tree for
2269 each directory and then using the B+ tree for searches.
2271 This feature is disabled by default. To enable it set
2272 the "BPlusTrees" registry value to 0x01 in the AFS Client
2273 Service Parameters key.
2275 * Correct a null pointer dereference in the symlink recursion
2276 detection code. (Introduced in 1.5.22)
2278 * Add the Network Identity Manager AFS Provider to AMD64
2283 * A new registry value, "BlockSize", can now be used to adjust
2284 the size of the internal buffers used to store file data.
2285 The default is 4KB. When set, the values 1..1024 specify
2286 multiples of 4KB. Values > 1024 specify actual block sizes
2287 and must be multiples of 4KB. The actual block size in KB
2288 can not be larger than the chunk size which defaults to 1MB.
2290 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
2293 * pts.exe and ptserver.exe built with -DSUPERGROUPS
2295 * Service Manager shutdown hints have been added. The hint
2296 specifies that it may take the AFS client service up to two
2297 minutes to shutdown.
2299 * Fix the recursion detection feature introduced in 1.5.22.
2300 The previous implementation broke the evaluation of paths
2301 using ../path or ./path. It also broke the evaluation of
2302 symlinks to absolute paths. For example, /afs/cell/path.
2304 * Fix an uninitialized variable in cm_MakeDir that could result
2305 in crash if the requested directory could not be created.
2307 * Fix a the VolumeStatusNotification log messages to save
2308 the input string in the circular log buffer so garbarge
2309 won't be output to afsd.log when the log is dumped.
2312 * Increase the SMB Maximum Path length to 260 characters
2313 from 256 as that is what CIFS supports.
2315 * On 64-bit Windows, use a unique description string for
2316 the 32-bit AFS Shell Extension.
2320 * In the AFS Admin Library, replaced all of the
2321 Ubik_Call(XXX_RPC,...) calls with Ubik_XXX_RPC(...). This
2322 allows parameter checking and proper type conversions to be
2323 performed. This is crucial when passing parameters that are
2324 64-bit wide on systems in which 'long' is a 32-bit value.
2325 Otherwise, the parameter list on the stack is interpretted
2326 incorrectly and the function not only doesn't succeed but it
2327 might corrupt memory and if lucky crash the application.
2329 The only program that we care about that uses this library
2330 is the AFS Server Manager. Now it doesn't crash.
2332 * Fixed smb directory searches for non-wildcard names. The
2333 optimized function should not have set the bulk in progress
2334 flag. This would adversely affect later directory searches
2335 that required a bulk search.
2337 * Detect infinite recursions (up to 512 stat cache objects
2338 in a chain.) If an infinite recursion is detected, return
2339 an error to stop the CIFS client from querying us forever.
2341 * Executables executed out of AFS name space can have their
2342 pages swapped out by Windows. Those pages will be reloaded
2343 when they are required. However, if the cache manager no
2344 longer has a callback and all of the file servers with
2345 the volume data are down, then the application will crash
2346 due to the inability to read the required page.
2348 We now track the down time of each server. If the callback
2349 expires after the server that issued it went down and all
2350 servers are down, we delay the expiration to permit the data
2351 in the cache to be continue to be used by Windows or an
2354 * The Windows AFS client has always had very poor performance
2355 with regards to directory management. In the case of
2356 directory updates caused by create, rename,and delete operations,
2357 the AFS cache manager will now perform local directory updates
2358 when possible in order to avoid re-reading the directory from
2359 the file server. This will be done in whenever both of the
2360 following conditions are true:
2362 - the data version on the directory increased by one as a
2363 result of the operation
2365 - all of the directory buffers necessary to make the change
2366 are present in the cache
2368 * The error message in the AFS Control Panel describing the
2369 need to be a Windows Administrator has been replaced with
2370 a message describing the need to be in the AFS Admins Group.
2372 * If the vldb server list is empty when attempting to evaluate
2373 a cell mountpoint it was possible for a crash to occur. This
2374 can only happen if there is an existing mount point and the
2375 is a failure attempting to load the server list from the
2376 CellServDB file or via AFSDB queries.
2378 * dirty buffers only have their dirty bytes stored to the file
2379 server instead of the entire chunksize
2381 * the default chunksize has been increased from 128KB to 1MB
2386 * changed the enum values for cm_serverRef_t state info to use a
2387 private name space to avoid collisions (srv_)
2389 * added a srv_deleted state for cm_serverRef_t objects. This
2390 state is set when cm_FreeServerList() is called with the
2391 CM_FREESERVERLIST_DELETE flag set. cm_FreeServerList() may
2392 not always delete the cm_serverRef_t from the list if it is
2393 still in use by another thread. the srv_deleted state means
2394 the object's contents are no longer valid and it must be
2395 skipped. It will be deleted the next time the object is
2396 freed and the refcount hits zero.
2398 * the srv_deleted state is also used when a file server reports
2399 either VNOVOL or VMOVED instead of marking the cm_serverRef_t
2400 as offline. This is done to prevent additional usage of the
2401 stale vldb data while waiting for the update volume request
2404 * added a state field to the cm_volume_t object (enum volstate
2405 vl_ name space) that maintains the state of the volume based
2406 upon the states of all of the cm_serverRef_t and cm_server_t
2409 * modified cm_UpdateVolume() to set the state of the cm_volume_t
2410 RW, RO, and BK to either vl_alldown or vl_online. There can't
2411 be any other states because cm_UpdateVolume() destroys any
2412 previous knowledge we might have had regarding busy or offline
2415 * modified cm_UpdateVolume() to update the volume name in the
2416 cm_volume_t to the volume base name if the previous value was
2419 * modified cm_FollowMountPoint() to check to see if the volume
2420 name is a volume ID and if so call cm_GetVolumeByID instead
2421 of cm_GetVolumeByName. This ensures that volume IDs are always
2422 looked up as numeric values. There is no longer a need to
2423 maintain a separate cm_volume_t containing the string representation
2426 * Added a flags parameter to cm_GetVolumeByName() and cm_GetVolumeByID().
2427 The first flag is a "CREATE" flag which is set by all existing
2428 calls. The flag is not set by calls to cm_GetVolumeByID() from
2429 the server probe code when volume status is being updated. We
2430 do not want the server probe operation to result in additional
2431 turnover in the cached volume data. The second flag is NO_LRU_UPDATE
2432 which is set when the server probe code updates the volume status.
2433 This flag will be used to prevent the server probe operation from
2434 changing the order of the least recently used queue.
2436 * Modified cm_GetVolumeByName to ensure that only one cm_volume_t is
2437 allocated for a given set of normal, readonly, and backup volumes
2438 regardless of whether or not the volume is accessed via name or
2439 ID number. The cm_volume_t namep field is always the base name
2442 * Added a new volume state, vl_unknown. This state is used as
2443 the initial state for all cm_volume_t when the cache manager starts,
2444 for each cm_volume_t at creation, and for each cm_volume_t when
2445 recycling. The cache manager does not know the state of all
2446 volumes in the world, only those that are in the cache and for
2447 which it has queried the VLDB and hosting file servers.
2449 * modified cm_GetVolumeByName() to initialize the state of a
2450 volume to vl_unknown. The actual state will be set when a
2451 cm_VolumeUpdate() call completes successfully.
2453 * changed name of scache hash table variables to avoid ambiguity
2454 when adding hash tables for volumes
2456 * fix a buffer overrun in sys\pioctl_nt.c pioctl().
2458 * modified cm_UpdateVolume() to handle the case in which there is
2459 no RW volume but there is are RO volumes for a given base name.
2460 This is done by querying for the ".readonly" volume name if the
2461 base name does not exist in the VLDB. We never query for the
2462 .backup name because under the current usage model a .backup
2463 volume may only exist on the server that the read-write volume
2464 is located. If there is no RW volume, there can be no .backup.
2466 * Added four hash tables for cm_volume_t objects to improve the
2467 search time of cm_GetVolumeByID(), cm_GetVolumeByName() and
2468 cm_ForceUpdateVolume(). One each for Name, RWID, ROID, and
2469 BKID. Three ID hash tables are necessary as long as it is
2470 desireable to maintain a single cm_volume_t containing all
2471 of the related RW, RO, and BK volume data. Having the RW and
2472 RO volume data in the same object is necessary for the
2473 implementation of cm_GetROVolumeID() which returns either the
2474 RO or RW ID depending upon the existence of RO volume instances.
2476 * Added a volume LRU queue so that volume reuse becomes fairer.
2477 This does not replace the all Volumes list which is used when
2478 it is desireable to walk a list of all the volumes whose order
2479 is not going to change out from underneath you which makes it
2480 safe to drop the cm_volumeLock.
2482 * handles volume hash table updates where volume name to
2483 volume ID number changes. The volume name remains
2484 constant in the cm_volume_t. if a vos rename is performed,
2485 the name of the volume will change and the volume IDs will be
2486 updated. Subsequent access to the old volume ID will create a
2487 new cm_volume_t with the new name.
2489 * Added a daemon thread operation to query the state of volumes
2490 listed as busy or offline. cm_CheckBusyVolumes() calls
2491 RXAFS_GetVolumeStatus() for each volume ID that is marked vl_busy
2492 or vl_offline. If the volume is now online, the status on the
2493 volume is updated. The default period is 600 seconds. This can
2494 be configured with the BusyVolumeCheckInterval registry value.
2496 * Added prototype for smb_IoctlPrepareRead() which was missing a
2497 return type in the function definition.
2499 * Added volume id lists to the cm_server_t. These lists are
2500 allocated in blocks of ~32 IDs. When a cm_PingServer()
2501 detects a change in server state, the state of the cm_volume_t
2504 * Added volID to the cm_serverRef_t object. volID is used
2505 to identify the volume for which the object is a referral.
2506 cm_FreeServerList() uses the volID to remove the volume
2507 from the cm_server_t.
2509 * In cm_Analyze, when VNOVOL or VMOVED are received,
2510 call cm_ForceVolumeUpdate() to force a refresh of the volume
2513 * Added cm_VolumeStatusNotification() which is used at the moment
2514 to log volume status changes to the trace log. It will also
2515 be used as the access point to the File System Filter driver
2516 notification engine.
2518 * Added an all cm_scache_t list to cm_data. This replaces the use
2519 of the stat cache LRU queue when we need to enumerate all
2520 entries. The LRU list order is not static and when using it to
2521 enumerate all entries it can result in items being missed or
2522 items being processed more than once.
2524 * Modified cm_Analyze(). Instead of reseting the busy or offline
2525 state of a volume and forcing a retry of the operation
2526 cm_Analyze will defer to the background daemon thread that will
2527 update the state once every 600 seconds.
2529 * Added the automatic generation of a Freelance ".root" read-write
2530 mountpoint that refers to the root.afs volume of the workstation
2531 cellname at the time the mountpoint is created.
2533 * re-write cm_Analyze to make better use of the known volume
2534 status. VL_Server queries cannot result in CM_ERROR_ALLOFFLINE
2537 * renamed cm_CheckBusyVolumes to cm_CheckOfflineVolumes.
2538 busy volumes will be reset to srv_non_busy by the function
2539 but there is no mechanism for querying the busy state other
2540 than by attempting to access the resource.
2542 * cm_Analyze will query the state of an offline volume before
2543 deciding whether or not to retry when all volume instances
2546 * the algorithm used after an IP address list change detection
2547 has been modified. we now check the state of the VLDB servers,
2548 then reset the rx connections, and finally check the state of
2549 the file servers. Previously we flushed the rx connections
2550 and then checked the servers in whatever random order they
2551 were first discovered.
2553 * fix the volume status reported by "fs examine"
2555 * various improvements to the AFS Network Identity Manager Provider.
2556 - Don't access freed objects after the window has been destroyed.
2557 - Write back the AFS cell list in the identity configuration
2558 even when its empty.
2559 - Disable all use of the provider when AFS is disabled for a
2560 particular identity.
2562 * Fix cm_cell_t leak. When evaluating a mountpoint that uses an
2563 alias instead of the full cellname, double check the full cell
2564 name and see if we have a matching cell before allocating a new
2567 * Dump more data with "fs memdump". All volumes and scaches and
2570 * Reverse the order of the all cells list. This ensures that
2571 the workstation cell is always first in the list.
2573 * If we recognize the server from which a callback revoke was
2574 received, then we can ensure that we only invalidate the status
2575 information for cm_scache_t objects belonging to that cell.
2577 * When building for amd64, use APPVER 5.02 as we do not support
2578 64-bit Windows 2000 and explicitly state the machine type in
2581 * no longer treat VMWare network adapters as equivalent to a
2582 loopback adapter. For some unknown reason, publishing the "AFS"
2583 netbios name on them frequently results in a name collision
2586 * Use read locks whenever possible instead of write locks when
2587 walking the allCellsp list. This permits better use of multiple
2590 * In 1.5.20 a bug was introduced by the fake status when PRSFS_READ
2591 is not present on the directory optimization. It would report
2592 mount points and directory objects for which there was no
2593 status information as a zero length file. Instead we can use
2594 the vnode number of the object to determine if it should be
2595 considered to be a directory or a file. If it is odd it is a
2596 directory; even it is a file.
2598 * Give Back Callbacks to the file servers in response to suspend
2599 and shutdown power management events. Only give up callbacks
2600 to UP servers and do so with a very short RX dead timeout.
2603 * Fix a race condition in the SMB FID cleanup code that could
2604 result in a stat cache mutex never being released.
2606 * Fix the graphics files that are installed as part of the
2607 IBM Administrators Guide.
2609 * Change the behavior of pioctl path parsing to not follow
2610 symlinks or mount points during evaluation. This results in
2611 the FID being the symlink or mount point and not the object
2612 they point to. This is consistent with the behavior of the
2615 * Enable AFSDB record lookup for command line tools such as
2618 * CellServDB update 2007-07-05
2620 * Added a Volume Status Plug-in Interface that permits third parties
2621 to receive notifications when the volume state changes occur.
2623 * Ensure that all writes to the afsd_alloc.log file use CR-LF.
2625 * Prevent the afsprocmgmt.dll from being unloaded while the
2626 background thread it starts is running.
2628 * Avoid a race condition when closing smb_fid_t objects.
2631 * fix install path of IBM Quick Start Guide when installed by
2634 * fix the client so that it can write into directories with
2635 only 'li' permissions.
2637 * fix installation of 32-bit Shell Extension for 64-bit systems.
2638 Install it under its own HKCR keys.
2640 * modify the wix installer to remove and upgrade install packages
2641 with the same version number.
2643 * when flushing volume location data every two hours or in
2644 response to "fs checkvolumes", also flush the mount point
2645 links in order to force their re-evaluation.
2647 * aklog was copying an uninitialized buffer as part of the
2648 foreign cell automatic pts registration
2650 * when obtaining directory listings, check for read permission
2651 on the directory. If read permission does not exist, fake the
2652 status information for the directory entries. This improves
2653 performance, prevents the client from pounding the file server,
2654 and prevents leakage of cached status information.
2656 * "fs getcacheparms" output was incorrect.
2658 * NetIdMgr AFS Provider should only automatically add token
2659 acquisition for cell foo.com if the realm of the cell matches
2660 the realm of the identity.
2662 * Fix a null pointer dereference in cm_MergeStatus()
2666 * network identity manager provider:
2668 - Correct for the fact that pr_CreateUser tampers with the
2669 contents of the username input buffer.
2671 - Take into account that the krb4 credential data structure
2672 does not have separate realm fields for the user and service.
2673 When krb524 is used with a cross-realm ticket, the realm
2674 describes the service not the user.
2676 * Many integrated logon changes:
2678 - Fix eventlog reporting. Do not attempt to log an event
2679 if the event source registration fails. Use DebugEvent0
2680 instead of DebugEvent when there are no parameters.
2682 - Modify the LOOKUPKEYCHAIN macro to recognize ERROR_MORE_DATA
2685 - Fix the reading of Domain specific configuration for
2686 LogonScript and TheseCells. Previously the dwSize value was
2687 being overwritten so that subsequent RegQueryValueEx call
2690 - Fix a memory leak in the TheseCells reading code.
2692 - Add support for Domain specific "Realm" specification. The
2693 realm is the realm to be appended to the username. When
2694 logging in as a domain or to the local machine, the specified
2695 "Domain" name is not going to be a valid realm name.
2697 - Construct a proper principal name based upon the domain
2698 specified realm for use in obtaining tokens with KFW.
2700 - If the domain specified "TheseCells" list includes the
2701 default cell, do not obtain tokens twice.
2703 * An unprotected call to the Kerberos v5 function,
2704 krb5_free_context(), can result in an invalid memory access
2705 exception if the contents of the krb5 profile, krb5.ini, cannot
2706 be parsed correctly. The error is located within the afskfw
2707 library which is used by afscreds.exe, afssvrmgr.exe, and
2708 afslogon.dll. When the error is triggered within afslogon.dll
2709 a blue screen condition can occur when the winlogon.exe system
2710 process terminates unexpectedly.
2714 * Do not stop the afs client service with an assert
2715 if the cell name cannot be determined when responding
2716 to a "fs whereis" command.
2718 * Integrated Logon (afslogon.dll). No longer attempt
2719 to preserve Kerberos 5 tickets for the logon session
2720 as that job is now supported by KFW 3.1 and above.
2722 * The Network Identity Manager plug-in will attempt to
2723 remove expired service tickets from the cache.
2727 * CellServDB update - GCO Public CellServDB 09 Mar 2007
2729 * AFS Server Manager and AFS Account Manager data cache
2730 moved from HKCU\Software\TransarcCorporation to
2731 HKCU\Software\OpenAFS to prevent crashes due to data
2734 * Reduce the default UP server probe time from 10 minutes
2735 to 5 minutes to match UNIX. This will not solve the NAT
2736 port mapping timeout issue because most NATs default to
2737 a value less that 5 minutes and this value ensures that
2738 the afs client server will start probes 5 minutes after
2739 the last probe set ended. The probe period for any one
2740 server is therefore longer than 5 minutes. The more
2741 servers in the list the longer the time between probes.
2743 * Before marking a server as down in cm_Analyze, retry the
2744 request using a new rx connection to the server. If the
2745 cache manager is behind a NAT, it is possible that the
2746 response from the server is being sent to the original
2747 address:port from which the rx connection was established.
2748 This will appear to the cache manager as the server being
2749 down when in fact it is not. Forcing the use of a new rx
2750 connection will result in the server responding to the
2751 most recent address:port assigned to the client by the NAT.
2753 If the server is in fact down, it will simply take a little
2754 longer to notice it.
2758 * Removing a directory that contained entries was failing with
2759 the wrong error code being returned to the CIFS client.
2763 * rxdebug exponential back-off. don't take down a server that
2764 is already struggling to respond.
2766 * Network Identity Manager plug-in registers help on the Help
2767 menu if it is supported by NIM. (KFW 3.2 and above)
2769 * Plug-n-play networking did not always work after coming back
2770 from a Windows Suspend. The AFS client service failed to
2771 re-compute the LAN adapter to bind to and instead attempted
2772 to bind to all LAN adapters. This would polute the public
2773 network with the name "AFS" which could lead to conflicts
2776 * Build installers using Microsoft Hotfixes provided at
2777 http://tinyurl.com/243svk for Visual Studio C RunTime Library
2778 Daylight Savings Time issues.
2780 * "fs checkserver" does not require admin privileges unless
2781 the probe interval is being changed.
2785 * A reference leak was discovered in the mountpoint traversal
2786 functionality which resulted in cm_volume_t objects not becoming
2787 eligible for re-use. This will trigger a panic condition when
2788 more than the maximum number of volumes is accessed.
2790 * Fixed the success test for the "fs memdump" operation. Success
2791 is now reported as such. Added cm_volume_t object information
2794 * A data corruption error has been fixed. Write requests
2795 queued for background processing were not retried upon failure.
2796 This leads to inconsistencies between the contents of the file
2797 in the local cache and the contents of the file on the file
2800 The implemented fix is to re-queue any request that has failed
2801 due to timeout, volume busy, volume offline, server busy, or
2802 server offline. Requests are not processed while all servers
2803 for a volume are inaccessible. When servers become accessible
2804 the background request will be retried.
2806 In addition, if a background write fails, do not mark the buffers
2807 as clean. Failure to preserve the dirty flag prevents the buffers
2808 from being written to the file server in the future.
2810 This fix does not address the possibility that the credentials
2811 necessary to perform a write request might expire between the
2812 time the request is queued and when the request can be processed
2813 due to server accessibility.
2815 * The Freelance volume does not support per user ACLs. The
2816 cm_HaveAccessRights function did not handle this special case.
2817 Requests for privileges greater than read|lookup would result
2818 in afsd_service.exe entering an infinite loop. ISOBuster is
2819 known to cause this behavior.
2821 * When deleting files, the SMB Server would attempt to obtain
2822 the access permissions for the parent directory instead of
2823 for the file itself. This would result in an incorrect
2824 assessment of whether or not the file can be deleted.
2826 * The Network Identity Manager AFS plug-in has had its string
2827 tables updated for consistency with the language used by NIM.
2828 All credentials are referred to as credentials instead of tokens.
2829 Drop-down combo boxes now in fact drop down.
2831 * The AFS Salvager was not properly being built on Windows.
2832 This is relevant only if the AFS servers are in use.
2834 * The default signal handlers for the AFS servers incorrectly
2835 processed SIGQUIT. Instead of stopping the server, an
2836 exception was raised.
2838 * Fix another volume refcount leak. This one was triggered by
2839 an inability to contact the vlservers in order to update the
2842 * Fix checklist control used by the drive mappings tabs in
2843 afscreds.exe and afs_control.exe.
2845 * Improve output of "fs memdump" related to buffer management.
2847 * Fix refcount leak of cm_buf_t objects when prefetching buffers
2848 already in the process of being fetched.
2850 * Improve cache performance by computing the buffer hash table
2851 size based upon the number of buffers so that the average
2852 number of entries in a bucket is seven.
2854 * Improve cache performance of read/write data by locally
2855 updating the dataVersion of all buffers associated with the
2856 file on each StoreData or StoreStatus operation provided that
2857 the dataVersion was incremented by one and the dataVersion of
2858 the buffer matched the original dataVersion of the cached
2859 status information for the file.
2861 * When recycling cm_scache_t objects, make sure that we
2862 clear all of the previous fields. It is no longer the
2863 case that we can assume that the stat cache object is being
2864 obtained only after a successful FetchStatus call.
2865 InlineBulkStatus will allocate stat cache objects for which
2866 the no status information is available.
2868 * When a file is being opened with the OPEN_ALWAYS disposition,
2869 do not require write permissions when the file already exists
2870 unless of course the file is being opened requesting write
2873 * When a file has been opened for read-only access, if the
2874 application requests an exclusive lock, obtain a read-lock
2875 instead of a write-lock. The purpose of the exclusive lock
2876 in this situation is to prevent changes to the file and the
2877 full file read lock will provide this guarrantee.
2882 * Enforce Unix Mode Write bit to prevent file deletion of read only
2885 * The CIFS READ and LIST permissions use the same value 0x0001.
2886 When a directory is being accessed, check for PRSFS_LOOKUP
2887 instead of PRSFS_READ.
2891 * Fix computations of short names of mount points. They were
2892 broken as part of the directory search optimization. (1.5.13a)
2894 * When all buffers are in use, periodically sleep the threads
2895 waiting for a buffer to become free while the global buffer
2896 lock is not being held in order to permit those threads that
2897 can free buffers to do so.
2899 * Power Management improvements. Maintain a global flag that
2900 specifies whether or not the service is in a suspend state.
2901 Do not panic if Netbios() returns NRC_BRIDGE meaning that the
2902 lana is no longer valid. Instead, stop the listener threads
2903 and if all listener threads are stopped, reset the lana_list.
2904 Allow the cm_Daemon() thread to periodically check the state
2905 of the smb listeners. If they are all stopped and the service
2906 is not suspended, attempt to restart them. If there are no valid
2907 lanas, return to the stopped state.
2909 * CreateX and NTCreateX use cm_CheckNTOpen() to test whether or not
2910 the user is permitted to obtain read or write locks. This function
2911 would obtain the lock and then drop it returning whether or not
2912 the lock could be obtained. If the lock was in fact required,
2913 CreateX/NTCreateX would then obtain it with cm_Lock(). The problem
2914 of course being that this pattern results in three RPCs to the
2915 file server (lock, unlock, lock). This is reduced to one RPC by
2916 implementing cm_CheckNTOpenDone() which frees the allocated byte
2917 range lock from cm_CheckNTOpen() after the cm_Lock() call is
2920 * Optimize the SMB FindFirst processing for the case in which there
2921 are no wildcards. If a cm_Lookup succeeds, generate a response
2922 from the stat cache entry we have a callback for. Its cheaper
2923 than walking each buffer of directory data associated with the
2924 directory stat cache entry when we have thousands of directory
2927 * Increase the SMB Ioctl MaxData size to support larger token sizes
2929 * Disable autorun of the AFS Server Wizard when the AFS Server is
2932 * Add AFS Server service dependencies on tcpip and PNP_TDI
2934 * Remove the AFS Server Configuration Tool shortcut
2936 * Add Freelance mode awareness to AFS Server Configuration Tool
2940 * Unlike previous versions of the OS, Vista performs a shutdown on
2941 the Microsoft Loopback adapter just like it would on a real adapter.
2942 This causes the smb_Listener threads to trigger a panic during a
2943 suspend/hibernate power event.
2945 The fix is to unbind from the network adapters in response to a
2946 suspend/hibernate power event and then rebind to the adapters
2947 when a resume power event is received. Note that the resume events
2948 are not reliably delivered so it is possible the afs service will
2949 not be accessible. However, this is the best we can do.
2951 * don't allow environment variables we set in afslogon.dll to be
2952 inherited by child processes
2954 * force an rx connection reset if VICECONNBAD or VICETOKENDEAD are
2957 * The old dirty buffer synchronization algorithm had a buf_IncrSyncer
2958 thread walking the all buffer list periodically searching for dirty
2959 buffers to write to the file server. This had several negative
2960 results. The algorithm ate up ever increasing amounts of CPU time
2961 even when AFS is idle as the size of the cache increases. Also,
2962 buffers were written to the file server in an order based upon the
2963 original buffer allocation which has nothing to do with the order
2964 in which the buffers became dirty.
2966 The new algorithm maintains a dirty buffer list. Items are added
2967 when the buffer is originally marked dirty. A buffer is only
2968 removed from the list by the buf_IncrSyncer when the buffer is no
2969 longer dirty. If the list is empty the thread goes back to sleep
2970 immediately without additional processing requirements.
2974 * Enable renaming of directories from the Explorer Shell when
2975 Offline Folders is enabled
2977 * An error is reported in response to the SMB ReceiveCoreFlush
2978 request when an error has not occurred.
2982 * optimize dirty buffer flushes in response to suspend operations
2983 do not flush if only loopback adapter is installed
2984 do no retry dirty buffer writes
2986 * do not reset netbios when performing loopback checks. doing
2987 so will break the existing lana bind
2989 * remove a deadlock in smb3.c QPathInfo
2993 * Add NetIDMgr AFS Plug-in
2995 * "Show Tray Icon" checkbox removed from afs_config.exe
2997 * Disable DNS Registrations for the Loopback Adapter and make sure
2998 that Netbios is turned on.
3000 * A reference count leak on stat cache objects was fixed
3002 * For debugging, tread ids of writers are now tracked on locks
3004 * Corrected an incorrect lock that was being held but not
3005 freed in a rare error case
3007 * Changed trace log end of line to CR-LF so it can be read
3010 * modify the buffer synchronization thread to write all dirty
3011 buffers and sleep only when there is no work to be done
3013 * improve behavior of cache manager when large numbers of
3014 temporary files are being created by multiple clients within
3017 * do not permit dirty buffers that belong to deleted files
3018 to be written to the file server. its a waste of bandwidth
3020 * do not leak stat cache references when the file server
3023 * afscreds: do not permit an attempt to obtain tokens if the
3024 afs service could not be started. this was causing stack
3025 overflows on Vista after resuming from a suspend
3029 * better handling of dot directory
3031 * fs sq and fs mkm now behave as on UNIX
3033 * add license to installers
3035 * Return Invalid Parameter error for unknown InfoLevels in
3036 SMB Tran2 Search Directory.
3038 * Correctly use WSAStartup and WSACleanup within the RX library
3040 * Check that afsd_service.exe is running before performing
3043 * Force AFS Shell Extension to load resource DLL using
3044 afs_shl_ext.dll as a base name.
3046 * (Wix) install afs_shl_ext_XXXX.dll and afs_creds_XXXX.dll in
3047 OpenAFS\Client\Program
3049 * Fix SMB_SET_FILE_END_OF_FILE_LENGTH
3051 * Do not panic if the maximum number of volume entries are in use
3052 and one of them can be recycled.
3056 * Fix the code that reads the backconnnectionhostnames value
3057 so that there is not an uninitialized variable
3061 * integrated logon could cause winlogin.exe to blue screen windows
3062 if it was called with a domain that does not exist on the machine.
3063 this is possible during a sysprep.
3065 * corrections to cm_buf_t handling to ensure that buffers are
3066 removed from queues before they are freed and that holds are
3067 obtained on the next buffer in a chain before releasing the
3070 * do not generate MessageBox dialogs on network errors. they
3071 cause more trouble then they are worth.
3073 * move GlobalAutoMapper into a separate thread so that it does
3074 not prevent service startup if the WNetAddConnection2 call
3077 * restrict hard dead and connection timeouts to the CIFS Session
3078 Timeout value. This is being done to try to prevent virtual
3079 circuits from being destroyed prematurely.
3081 * Fix the code that writes the backconnectionhostnames value
3082 to ensure that the data buffer is written with the correct
3085 (there were no windows specific changes between 1.5.4 and 1.5.5)
3089 * Corrections to the Integrated Logon functionality that permits
3090 Kerberos 5 TGTs to be communicated into the user session. The
3091 transfer is now more secure and succeeds when the user does not
3092 have read permission to the %SystemRoot%\TEMP directory.
3094 * NSIS installer updated to version 2.18
3096 * WiX installer updated to version 2.0.4221.0
3098 * Updates to the IFS functionality for 64-bit correctness and
3099 permit compilation. The IFS functionality works for pioctl()
3100 operations but not much else at this point.
3102 * A major re-write of the OpenAFS implementation of the CIFS Tran2
3103 query and set functions.
3105 * Ignore the Windows CIFS client session timeout entirely. Timeout
3106 only after the AFS Hard Dead Timeout period expires.
3110 * move headers, libraries, and samples from OpenAFS\Client to
3113 * remove the auto-registration of AFS IDs for foreign cells
3114 from integrated logon. this permits afslogon.dll to avoid
3115 initializing the RX library. afslogon.dll is a network
3116 provider library and as such it is loaded into all applications
3117 which perform a CreateProcessAsUser call. Initializing the
3118 RX library spawns background threads to listen for inbound
3119 packets. When the network provider is unloaded from the
3120 process, the RX library does not cleanup the threads. If they
3121 continue to execute after the library is unloaded, the process
3124 * implement inline bulk status calls. this will significantly
3125 improve the performance of directory listings.
3127 * prevent a crash when evaluating mountpoints to volumes that do not
3130 * new command: fs uuid [-generate]
3131 displays the current uuid and allows a new one to be generated
3132 on the fly without restarting the afs client service.
3134 * fix the 64-bit installers to properly document the install
3137 * move help files to the proper location so they can be located by
3138 the system tray tool and the control panel.
3140 * "over quota" and "disk full" errors were not being reported to the
3143 * largefile support has been added. Files greater than 31-bits can no
3144 longer be written to file servers which do not support large files.
3145 When file servers support large files, files up to 64-bits can be
3148 * cm_SyncOp could return with the input buffer unlocked upon error
3149 which would result in a panic condition. This bug was introduced
3152 * vlserver failover when freelance mode did not work. This prevented
3153 the root.afs volume from being loaded if the first vlserver could
3158 * A bug in the CIFS Standard Info response packet caused directories
3159 to appear as standard files. This prevented Microsoft Vista from
3160 being able to walk the AFS directory tree.
3162 * Mount points must be reported as directories in response to
3163 Query Path Info operations.
3165 * A deadlock was detected when performing "fs flushall" if the
3166 file server reports VNOVNODE. The scp->createBufferLock is
3167 already held by the current thread and there is no mechanism
3168 to propagate the knowledge. Therefore, an alternate mechanism
3169 for clearing the cache must be developed. A new function
3170 cm_RecycleSCache(scp,flags) has been extracted from
3171 cm_GetNewSCache(). This function performs the task of recycling
3172 an cm_scache_t object. When called from cm_GetNewSCache() with
3173 no flags the expectation is that there are no associated buffers
3174 that are queued to be read or written. When called from
3175 cm_Analyze() with the CM_SCACHE_RECYCLEFLAG_DESTROY_BUFFERS
3176 flag, any queued buffers will be de-queued and marked as if
3177 the operations were performed so the data can be discarded.
3179 * CM_ERROR codes were logged as "unknown" by cm_Analyze.
3181 * In response to VNOVNODE, the parent is only discarded if the current
3182 cm_scache_t is not a directory.
3184 * In the Ioctl FlushFile and FlushVolume functions, there were no
3185 checks to protect against flushing the Freelance SCache entries.
3187 * In FlushFile, the wrong cm_scache_t object was being released.
3189 * In cm_GetNewSCache, do not allow Freelance SCache entries to be
3190 recycled. Choose a new entry if cm_RecycleSCache fails.
3192 * Speed up the performance of the cache manager by not
3193 holding the smb_fid_t mutex across calls to cm_SyncOp
3196 * Ensure that all smb_fid_t flag references are protected.
3198 * Remove a deadlock between smb_fid_t mutex and
3199 smb_rctLock introduced in 1.4.1.
3201 * Fix aklog AFSID lookup to use not send the realm
3202 name when the realm is the local realm for the
3203 cell. The PT Server doesn't strip the realm
3204 name when it is the local realm.
3206 * Treat "\\afs\*." as an alias for "\\afs\all"
3208 * Dynamically adjust the priority of server threads
3209 based upon the age of the CIFS request that is
3210 being processed. The priority is bumped one level
3211 for every 15 seconds of processing time.
3213 * The original openafs contribution mis-used the
3214 cm_GetCallback function. This function should only
3215 be called as a side effect of a call to
3216 cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK). The way it
3217 was being called results in a FetchStatus RPC being
3218 performed even when the client already has a valid
3219 callback and does so in a manner that destroys the
3220 synchronization of the threads that are calling
3221 cm_SyncOp. If nothing else this results in a
3222 significant performance penalty.
3224 * More changes to cleanup of smb_vc_t objects in
3225 order to prevent race conditions.
3227 * Reduce the amount of Power Management event logging
3230 * If the client detects that the IP address has
3231 changed, force the use of new RX connections. This
3232 enables the file server to recognize that the client
3235 * Remove a race condition and deadlock associated
3236 with cleanup of smb_vc_t objects.
3238 * Fix the ANSI filename support. One function that
3239 required it was missing the translations.
3241 Changes not present in 1.4.1 and since 1.5.0: 1.5.1 released
3244 * Add a DllMain() function to afspthread.dll. Perform
3245 cleanup of thread local storage, rx queues, critical
3246 sections, event handles, etc. This was implemented
3247 to avoid crashes in afspthread.dll when applications
3248 which load/unload the dll attempt to terminate.
3250 * Lock request behavior has been altered. As of this
3251 release a locally allocated lock will be used whenever
3252 the volume is RO or the requesting user's permissions
3253 are 'rl'. This will allow end users to execute programs
3254 from RW volumes when they do not have the 'k' privilege.
3256 Note that a recent discovery indicates that the 'k' bit
3257 is only supposed to control the ability to obtain read
3258 locks. Write locks are to be controlled by the 'w' and
3261 * Code has been added to enable the cache manager to be
3262 able to obtain the AFSID of the authenticated user.
3263 This is required to be able to determine when "Insert"
3264 privileges should apply. AFS considers a file to be
3265 a new file whenever the owner's AFSID matches the
3266 AFSID of the user. The current implementation relies
3267 on the name specified during the SetToken call to be
3268 correct. This will be replaced in the future with a
3269 new RPC that allows the AFSID of the authenticated
3270 user to be obtained.
3272 * A 32-bit-tools installer for use on 64-bit systems
3273 is now available. These tools are for use with 32-bit
3274 versions of Kerberos for Windows or JPSoftware's 4NT.
3276 Since 1.4.1 rc10: 1.4.1 final released (9 April 2006)
3278 * Remove the use of the "active_vcp" variable. This variable
3279 no longer serves any useful purpose and was the cause of
3280 race conditions that produced under counts of the smb_vc_t
3283 * Change default minidump output to include data segments
3285 * restore access to cm_IoctlGetVolumeStatus and
3286 cm_IoctlSetVolumeStatus calls for AFS pioctl operations.
3287 Still need to figure out why they are being called
3288 for SMB/CIFS IOCTL operations instead of Trans2GetFileInfo.
3290 Changes not present in 1.4.1: 1.5.0 released (16 Feb 2006)
3292 * Update build system to include the "sys_name" in the
3295 * optimize the number of calls to obtain the smb_user_t for a
3296 smb virtual circuit.
3298 * add logging of server addresses, status and capabilities when
3301 * prevent multiple threads from pinging the same server simultaneously.
3302 instead, threads block and then use the result of the ping that was
3303 already taking place.
3305 * Add registry values to allow the defaults for daemon checks to be altered
3306 at startup. Log values to afsd_init.log.
3308 daemonCheckDownInterval
3309 daemonCheckUpInterval
3310 daemonCheckVolInterval
3311 daemonCheckCBInterval
3312 daemonCheckLockInterval
3313 daemonCheckTokenInterval
3315 * Add EventLog messages for transitions in fs crypt state.
3317 * Byte range locking (when locks will be obtained from the
3318 is still under discussion)
3320 * 64-bit Windows operating systems now supported
3321 (XP SP2 64-bit, 2003 SP1 64-bit, 2003 R2 64-bit)
3322 This is a native 64-bit implementation of OpenAFS for Windows.
3323 There is currently no 64-bit version of KFW and therefore
3324 there is no (easy) method for obtaining AFS credentials.
3325 32-bit KFW 3.0 can be used with 32-bit OAFW libraries
3326 to obtain AFS tokens but there is not for the feint of
3329 * A faster implementation of the fcrypt algorithms improves
3332 * The afsd_service.exe now supplies Event Log Message Text.
3333 The event log no longer generates warning text due to lack
3334 of message configuration.
3336 * When afs servers are unreachable and network stack returns
3337 a WSAEHOSTUNREACHABLE error, the AFS client will immediately
3338 mark the attempted server as being down. This allows for
3339 very fast response to the calling application. This change
3340 has a pronounced improvement when the Windows Explorer is
3341 used on disconnected mobile computers.
3343 * Added DebugOutputString debugging of the RX library
3345 Since 1.4.1 rc9: 1.4.1 rc10 released (10 March 2006)
3347 * Fix the rx-lwp implementation to randomly generate rx call
3348 identifiers which are supposed to be unique at a given point
3349 in time. This affects some of the command line tools used
3350 for debugging but not the AFS Client Service itself which
3351 uses the rx-pthreads implementation.
3353 * xstats_fs_test.exe and xstats_cm_test.exe are now built for
3354 Windows although they are not distributed
3356 * don't generate an exception when logging unexpected SMB packets
3358 * the data structures used by cm_IoctlGetVolumeStatus and
3359 cm_IoctlSetVolumeStatus are not compatible with the CIFS
3360 Get/Set File Information calls so don't bother executing them.
3364 * Fix data written to registry as part of BackConnectionHostnames
3367 * Fix initialization bug when allocating cm_server_t objects that
3368 would cause servers to appear to be down after each allocation
3369 or renewal. Server would appear up after the next check servers
3374 * Windows will return WSAECONNRESET to the next receive packet
3375 operation on a socket if an ICMP packet is received in response
3376 to a packet that was sent and could not be delivered because
3377 the port was unreachable. We have to treat this as
3378 WSAEWOULDBLOCK so that we can retry the select() check and
3379 not mark the connection as bad.
3381 Since 1.4.1 rc6: 1.4.1 rc7 released (14 Feb 2006)
3383 * The client manager maintains a least recently used list of
3384 all the ACLs. If an ACL is already the most recently used
3385 do not bother to remove it from the list and place it on the
3388 * As part of the process of adding a new server, set the
3389 server is down flag and then perform a ping. If the server
3390 is up, the flag will be reset. If the server is down, the
3391 client will not timeout unnecessarily while attempting to
3394 * If a Mountpoint String is the empty string, return PATH NOT
3397 * Modify behavior of the Network Provider to avoid querying the
3398 profile location if integrated logon is known to be disabled.
3400 * Correct behavior of 'LogoffPreserveTokens'. When set, tokens
3401 will always be preserved regardless of whether or not integrated
3402 logon was used to obtain tokens. When unset, tokens will be
3403 destroyed regardless of whether integrated logon was used to
3404 obtain the tokens unless the profile is stored in AFS.
3406 * Disable LogoffTransferToken functionality. The SMB V3 Logoff
3407 message is being delivered even when the user is not logging
3408 off of the computer.
3410 * Fix a deadlock condition introduced in rc6.
3412 * Cell names are case-insensitive. Since the founding of OpenAFS
3413 there has been a bug that treated "cell", "Cell", and "CELL" as
3414 different cells. This would result in what appeared to be
3415 random disappearance of user tokens or a failure to authenticate
3416 even when tokens were listed by the "tokens" command.
3418 Since 1.4.1 rc5: 1.4.1 rc6 released (2 Feb 2006)
3420 * Additional auditing of the smb code revealed that smb ioctl
3421 operations were failing to free smb_fid_t objects
3423 * The File Identifiers assigned to smb_fid_t objects could be
3424 given the value 65535 which is 0xFFFF which is also the
3425 INVALID_HANDLE value. Both 0 and 65535 are now treated as
3428 * Audited all assignments to fields and ensured that they are
3429 protected by obtaining the relevant mutex object.
3431 * Re-wrote the VNOVNODE handling to avoid a potential deadlock
3432 if both the current cm_scache_t object and its parent are
3433 both returning VNOVNODE.
3435 * Removed all the logic associated with the 'dead_vcp' variable
3436 which must have been added for debugging because its used
3437 was never thread safe and it truly served no purpose.
3439 Since 1.4.1 rc4: 1.4.1 rc5 released (25 Jan 2006)
3441 * Re-enabled LogoffTokenTransfer and LogoffTokenTransferTimeout
3442 registry settings. Default is ON and 120 seconds. If set to
3443 OFF, tokens will not be destroyed.
3445 * Fixed a number of race conditions and reference count problems
3446 associated with global queues.
3448 * Added a method to detect abandoned smb virtual circuits so their
3449 associated file handles, locks, user tokens, and smb session
3450 resources can be released.
3452 * Added a mechanism to allow user tokens to be preserved
3453 during integrated login and be freed after logoff.
3455 * Check access permissions to the file before the file is
3458 * More improvements to the handling of SMB Virtual Circuits
3459 when SMB sessions are closed.
3461 * Correct a race condition introduced in rc4 that could result
3462 in a crash of afsd_service.exe.
3464 Since 1.4.1 rc3: 1.4.1 rc4 released (11 Jan 2006)
3466 * Properly handle pre-mature SMB session closures.
3467 File handles, tree connections, and user sessions are now
3468 freed when the virtual circuit is closed.
3470 * smb_username_t and smb_user_t objects are now properly reference
3471 counted and freed as appropriate.
3473 * disable to use of AFS file locks. byte range locking is still
3474 implemented but all locks are managed locally. This prevents
3475 multiple processes from editing the same file on the same
3476 machine but it does prevent multiple processes on separate
3477 machines from modifying the file.
3479 Since 1.4.1 rc2: 1.4.1 RC3 released (4 Jan 2006)
3481 * A fix to "fs setserverprefs -vlserver". Multiple calls with
3482 the same server parameter would result in a panic condition
3483 if the server had not already been contacted by the AFS
3486 * Better warnings are written to the afsd.log file when
3487 the Windows RPC Protocols are not properly configured as
3488 is often the case on Windows 2000.
3490 * fix the procmgmt library to restore C RTL signal handlers when
3491 it is being detached from the process prior to process termination.
3493 Since 1.4.1 rc1: 1.4.1 RC2 released (30 Nov 2005)
3495 * close a lock leak in the rx library that could be triggered while
3498 * prevent removal of drive mappings in response to afscreds.exe -M
3500 * add logic to process VNOVNODE in cm_Analyze. Force re-evaluation of symlink
3501 strings and flush the stat cache entry.
3503 * prevent server objects from being freed if user preferences are set.
3505 * Update WIX support to 2.0.2419.0
3507 * Release all locks on the server at service shutdown.
3509 * A down server should be marked up as long as it responds to the client.
3510 It doesn't matter if the server returned an error as long as the error
3511 is something other than VRESTARTING. If there is such an error, all
3512 connections on that server should be marked for replacement.
3514 * Change the check "up" servers period from once per hour to once every
3515 ten minutes. This matches the behavior of the Unix cache manager.
3517 Post 1.4.0 revisions: 1.4.1 RC1 released (21 Oct 2005)
3518 * Byte Range Locking support has been added. Architectural documentation
3519 is located within src/WINNT/afsd/cm_vnodeops.c. Best efforts are
3520 made to enforce Windows mandatory locking model using locally managed
3521 byte range locks with full file locks obtained from the file server.
3522 Read-only volumes provide the user with an implied locking privilege.
3523 Read-write volumes require that a user be granted the locking privilege
3524 in order for a lock to be obtained. As most Windows applications use
3525 Shared file access modes, they require the ability to obtain locks in
3526 order to access files.
3528 * Implement error translation functions to convert Universal Error Codes
3529 output by modern servers to local errno and winsock error values.
3530 This fixes the error message problems using pioctl() calls from fs.exe
3531 and the AFS Shell Extension.
3533 Since 1.4.0 rc8: 1.4.0 final tagged (20 Oct 2005); released (31 Oct 2005)
3534 * Allow the AFS Client Service to set the firewall configuration by
3535 retrying until after Windows determines the boot period is over.
3537 * Do not attempt to configure the firewall on Windows 2000, XP pre-SP2,
3540 * Detect IP address changes from within the AFS Client Service for the
3541 purpose of checking down servers sooner
3543 * If the file server or vldb server report RXKADUNKNOWNKEY, return
3544 SEC_E_NO_KERB_KEY to Windows. This provides an indication to the
3545 user that there is a Kerberos key problem. Previously the user would
3546 receive no feedback when there was something wrong.
3548 * Fix the large integer library support to properly handle values greater
3551 Since 1.4.0 rc7: 1.4.0 rc8 released (15 Oct 2005)
3552 * Updated CellServDB
3554 * Fixed file name pattern matching. Question marks can match nothing
3555 if found at the end of a component.
3557 Since 1.4.0 rc6: 1.4.0 rc7 released (8 Oct 2005)
3558 * Kerberos 5 ticket lifetimes should adhere to KFW Leash lifetimes
3560 * The MSI installer now has a new Product ID for each build. This
3561 is to allow for automated upgrades. The Upgrade ID is remaining
3564 * Volumes that were cached and then subsequently were salvaged,
3565 moved or otherwise taken temporarily offline (as opposed to being
3566 busy) would become unreachable. A distinction is now made between
3567 a volume being offline and a server being down.
3569 Since 1.4.0 rc5: 1.4.0 rc6 released (2 Oct 2005)
3571 * Fixed several AFS Shell Extension issues:
3572 - Checking Server Status would only work if a specific cell was
3574 - If servers were down, the Server Down dialog would write to
3575 invalid memory and cause explorer.exe to crash
3576 - Default Push Buttons were not specified for many dialogs
3577 - The extension would be unloaded by the Explorer Shell and
3578 never be reloaded under some circumstances.
3580 * Fixed the AFS System Tray menu to disappear if another Window
3581 was selected before a menu item was selected
3583 * Added volume owner and group information to "fs examine" output
3585 * Added fs and registry support for enabling/disabling/clearing
3586 rx statistics gathering.
3588 Since 1.4.0 rc4: 1.4.0 rc5 released (25 Sep 2005)
3590 * New Product Code GUIDs issues for 1.4 and 1.5 branches. Separate
3591 GUIDs for each platform.
3593 * Documentation directory structure and default installation rules
3596 * New OpenAFS for Windows HTML Release Notes installed
3597 as part of the documentation set.
3599 * Add thread id to rx debugging messages
3601 * When all servers are offline, return a bad network path error
3602 immediately. This is necessary to allow Explorer to do the
3603 right thing and not hang. The background daemon thread will
3604 check the down servers every 30 seconds to see if they are
3607 Since 1.4.0 rc3: 1.4.0 rc4 released (14 Sep 2005)
3609 * The AFSCache file has been modified to store both the
3610 serial number of the volume on which the file resides
3611 and the SID of the local machine. If either change,
3612 a new UUID is generated.
3614 * add "fs flushall" command that forces all data buffers
3617 Since 1.4.0 rc2: 1.4.0 rc3 released (4 Sep 2005)
3619 * add support for '/' instead of '\' in pioctl() calls
3621 * Apply AFS Client Admins group protect to AFS Shell Extension
3623 * Add support for \\afs\<name> to most AFS Shell Extension
3624 commands except the symlink methods.
3626 * when installing the Microsoft Loopback Adapter, enable
3627 MS Client for Networks and prevent an install failure
3628 by not calling CoInitialize twice in the same thread.
3630 * reload cell vldb values from the CellServDB every two
3631 hours in case it changes
3633 * When updating cell information from DNS, be sure to set
3636 * Add support to allow use of \\AFS\<foo> where <foo> is
3637 either a mount point or symlink. As <foo> is normally
3638 treated as a share name, we transform it into \\AFS\all\<foo>
3641 * Init 'code' to prevent false errors when integrated logon
3642 is disabled and the service is not running
3644 Since 1.4.0 rc1: 1.4.0 rc2 released (28 Aug 2005)
3646 * Removed trace log messages that were hampering performance.
3648 * Fixed a deadlock that was being triggered by editing Word
3649 documents stored within AFS with WinWord 2003.
3651 * Bit 3 of the TraceOption registry value is now used to set the
3652 default for "fs trace" in the non-Debug builds. The new default
3653 is off for release builds and on for debug builds.
3655 * Bit 2 of the TraceOption registry value can be used to turn on
3656 real-time output of debug log entries to the Windows Debug
3657 Output monitor interface. This data can be viewed with tools
3658 such as Sysinternal's DbgView.
3660 Since 1.3.87: 1.4.0 rc1 released (17 Aug 2005)
3661 * When the cache manager reports ALLBUSY or ALLOFFLINE for
3662 a resource we should not be returning status codes such
3663 as Network Busy or Remote Host Down because these errors
3664 will imply that the AFS Client Service is Busy or Down
3665 and that is simply not the case. Instead we will return
3666 Bad Network Path as the path is temporarily not available.
3668 Instead of returning Sharing Paused when there is a
3669 timeout error, return Timeout. Once again, the AFS Client
3670 Service is not paused.
3672 * afscreds.exe would display an Obtain Creds dialog when
3673 the expired credentials reminder was triggered even
3674 if there was no network path to the KDC. This is prevented
3675 by adding KDC probe logic to the reminder thread.
3677 * afscreds.exe would display expired tokens no differently
3678 than unexpired ones. This would make it difficult for the
3679 user to distiguish when the tokens were expired. For the
3680 English build added a new resource string "(expired) that
3681 is displayed instead of the expiration time.
3684 * "fs wscell" when executed in freelance mode will return the
3685 name of the cell configured in the registry. The root.afs
3686 volume is not loaded from this cell, but it is used for the
3687 default for aklog and integrated login.
3689 * "fs mkmount, fs rmmount, symlink make, symlink remove" will
3690 no longer work on \\AFS\all when freelance is being used
3691 unless the user is a member of the "AFS Client Admins" Windows
3694 * some more checks performed during persistent cache validation.
3697 * Optimize calls to cm_CheckServer from cm_Analyze. Only check
3698 down servers for the current cell not all cells.
3700 * If the scache LRU list becomes corrupted, the AFS Client Service
3701 might enter an infinite loop at startup. Detect the corruption
3702 and discard the cache contents in this case.
3704 * Fixed the Explorer Shell's remove mountpoint functionality.
3706 * Fixed a deadlock caused by the holder of cm_aclLock attempting to
3707 obtain a mutex lock on a cm_scache_t object whose lock is held by
3708 a thread attempting to obtain the cm_aclLock.
3711 * Really fixed DNS AFSDB queries to ensure that "csail" cannot be
3712 misinterpretted as "csail.mit.edu" when the resolver
3713 is configured to append ".mit.edu" to failed DNS queries.
3715 * Added a new registry key, "LogoffPreserveTokens" (see registry.txt),
3716 that can be used to force the preservation of user tokens upon logout.
3718 * Update the NSIS install scripts to use NSIS 2.07. This release adds
3719 recommended installation categories: AFS Client, AFS Administrator,
3720 AFS Server, AFS Development Kit. Each category includes a different
3721 default subset of the OpenAFS components.
3723 The OpenAFS logo is now associated with the NSIS uninstall entry
3724 in the Add/Remove Programs control panel.
3726 * The user name associated with AFS tokens when obtained with
3727 integrated login, the afs systray tool, or aklog will always
3728 include the full Kerberos 5 user principal name regardless of
3729 whether or not the cell is local to the realm.
3731 * Modify integrated login so that it does not enter an infinite
3732 loop if the service is not set to auto start.
3734 * Added asetkey.exe used to set a Kerberos 5 key for use by the
3737 * Added uninstall.exe to Wix installation
3739 * More modifications to algorithms used to wake sleeping threads.
3743 * Changes to the algorithms used to wake threads when they are
3744 put to sleep because of conflicts over buffer operations.
3747 * The OpenAFS integrated logon obtains Kerberos 5 tickets as part of
3748 the process of producing AFS tokens. The tokens are stored within
3749 the AFS Client Service but the Kerberos 5 tickets have been discarded.
3750 New functionality has been added to temporarily copy the tickets to
3751 a file ccache during the short window between the NPLogonNotify and
3752 WinLogon Logon Event. The Logon Event handler starts a process as
3753 user, afscpcc.exe, that copies the tickets into the default ccache.
3754 The file ccache is then destroyed by the event handler.
3756 * During synchronization operations on stat cache entries, if threads
3757 are forced to sleep it was possible that the threads would not be
3758 woken under heavy load.
3760 * Enforce the delete on close flag specified by the NTCreateX()
3763 * A race condition in the rx library was fixed that resulted in
3764 a deadlock in rx_NewCall when the RX_CONN_MAKECALL_WAITING flag
3765 was cleared when threads were still waiting to use the
3766 connection. Also, fixed a potential case in which the
3767 connection could be destroyed while threads were waiting to use it.
3769 * The DNLC freelist has been seen to become corrupted with still
3770 active entries being stored on the freelist. Changes were made
3771 to perform a better job of cleaning entries before placing them
3772 onto the freelist; marking them with the correct magic value when
3773 purging the list; and allowing the list to be purged on startup
3774 if the cache validation fails.
3776 * Windows Crash Reporting does not get triggered for afsd_service.exe
3777 because it provides its own top level exception handler. This patch
3778 places an explicit call to ReportCrash() on platforms which support
3781 If you configure Crash Reporting via AD Policy to capture crash
3782 reports within the domain, then you will receive this output.
3784 * Generate MiniDumps independent of the Windows Crash Reporting.
3786 (a) If an exception occurs,a minidump will be generated at the path
3787 %TEMP%\afsd.dmp. The type of minidump defaults to Normal but
3788 can be set to any other type via use of the
3791 [HKLM\SOFTWARE\OpenAFS\Client] MiniDumpType (DWORD)
3793 (b) MiniDumps can be generated at any time using the "fs minidump"
3794 command. This allows you to generate a minidump without
3795 stopping the process or even requiring the use of a debugger
3798 * an RX connection which reports bad ticket now treats it the same way
3799 the expired ticket is treated. The ticket is removed from the
3800 connection and a new anonymous connection is established.
3802 * fs memdump now requires that the caller be in the AFS Client Admins
3805 * additional debugging information is provided in cm_Analyze when the
3806 error code is ignored.
3808 * Fix a regression introduced into the pioctl() client code which
3809 will cause a crash in winlogon.exe, explorer.exe and fs.exe if
3810 MIT KFW is not installed. (1.3.82a)
3812 * Remove AFS Gateway option from OpenAFS Control Panel (1.3.82a)
3815 * Fix a race condition in cm_EndCallbackGrantingCall() that
3816 could leave 'nrevp' pointing to freed memory. In the process,
3817 optimize the processing to ensure that the applications monitoring
3818 the status of the stat cache entry are only notified at most
3821 * Conditionalize access to the Explorer Shell AFS->Mount Point
3822 ->Remove menu item. Disable the item if the selected item(s)
3823 are not mount points.
3825 * Activate AFS RPC (RX) Free Packet Queues stored in Thread
3826 Local Storage. This eliminates thread contention on the
3827 global RX free packet queue. This should improve performance
3828 on multi-CPU systems.
3830 * Fix the IP address filtering code to properly load/unload
3831 the IPHelper DLL so that Windows XP/2003 and beyond systems
3832 do not have to manually probe the registry.
3834 * CleanupACLEnt() was not being called consistently with
3835 the cm_scache_t object referred to by the back pointer
3836 mutex locked. This could in very rare conditions lead to
3837 an invalid memory access.
3839 * Added a script command to the msi installer "afs_replace"
3840 which can be used to replace one version of OpenAFS with
3841 another without requiring an uninstall and reinstall if
3842 the installed msi is no longer in its original location.
3844 * Apparently the problem with multi-domain forests with cross-
3845 realm trusts to non-Windows realms was not entirely solved.
3846 The authentication to the AFS SMB service failed because
3847 the wrong name was being used. Using ASU as an example,
3848 the authentication was being performed with the name
3849 "QAAD\user" (an account in the forest root) and not
3850 "user@ASU.EDU (the MIT Kerberos principal used to login with)
3852 The solution was to add an additional dependency on KFW
3853 in order or to be able to easily obtain the client principal
3854 name stored in the MSLSA ccache TGT. This information is
3855 used in two locations:
3857 - the pioctl() function
3859 - a new WinLogon Event Handler for the "logon" event.
3861 The pioctl function will now be able to use the correct
3862 name when calling WNetAddConnection2() and the "logon"
3863 event handler will now be able to call WNetAddConnection2().
3864 The hope is that the "logon" event handler will be called
3865 before the profile is loaded but I have not guarrantee
3869 * Fixed a locking error in cm_TryBulkStat() which had the
3870 potential of deadlocking the system for the length of time
3871 it takes to perform a bulk status fetch operation.
3873 * Replaced time conversion code (UnixTime <-> FILETIME) to
3874 be completely arithmetic instead of relying on a bizarre
3875 algorithm involving a variety of C RTL time functions.
3876 This has the side effect that UnixTime and FILETIME which
3877 are both stored in UTC are interpretted as UTC throughout
3878 the year. Windows will apply the same localization to AFS
3879 as it does to NTFS. Applications which rely on the ability
3880 to sync files between the two file systems will no longer
3881 see the timestamps of files in AFS change an hour relative
3882 to the files stored in NTFS or Windows based backup devices.
3884 * Fixed a invalid memory access under a bizarre circumstance.
3885 Windows will allow a physical mass media device to be
3886 installed via Plug N Play to the system and will assign it
3887 the lowest available drive letter. This is true even if the
3888 drive letter is currently assigned in the user session to
3889 a network device via NET USE (or its equivalent.) When this
3890 happens, queries sent to the CIFS server will contain invalid
3891 data. This invalid data was not being caught by the AFS
3892 Client Service and was resulting in a crash when Freelance
3895 * Fixed a reference count error when registering callbacks
3896 on a stat cache object if the callback was already registered
3898 * Add a case to cm_Analyze() in order to handle the case where
3899 cm_GetConnByMServers() returns CM_ERROR_NOSUCHVOLUME because
3900 the server list for the volume is empty. In this case, force
3901 an update of the volume info and retry.
3903 * Insert a missing cm_EndCallbackGrantingCall() which could
3904 result in threads waiting for a callback to complete to never
3907 * In the persistent cache, there is a maximum number of volume
3908 entries. Allow volume entries to be reused if the maximum
3909 number have been allocated and their reference count is zero.
3911 * If we already have a dead virtual connection object, don't
3912 ignore the fact that additional dead vc objects must be taken
3915 * Removed a deadlock condition introduced in the summer of 2004.
3916 Do not hold mx locks around calls to RX functions including
3917 rx_NewCall(). That is what reference counts on the rx_connection
3920 * Fixed an initialization error in afslogon.dll which could
3921 result in random behavior including a failure to terminate
3922 the NPLogon function. (1.3.80b)
3924 * Fixed an error preventing the use of SMB authentication
3925 on Windows 2000. (1.3.80a)
3928 * Updated CellServDB to Public CellServDB 16 Mar 2005
3930 * Fixed DNS AFSDB queries to ensure that "csail" cannot be
3931 misinterpretted as "csail.mit.edu" when the resolver
3932 is configured to append ".mit.edu" to failed DNS queries.
3934 * Fixed another case in which the client would replace connections
3935 to the host when it was not necessary. In this case the
3936 situation would occur if cryptall was on and the connections
3937 were unauthenticated due to lack of an appropriate token.
3939 * OpenAFS for Windows has failed to work at sites which are
3940 utilizing a cross-realm trust between an MIT/Heimdal realm
3941 and a multi-domain Windows forest when the workstation being
3942 accessed is not located in the root domain. This is caused
3943 by a bug in the workstation which was triggered after the
3944 introduction of Windows 2003 Server. When the bug is triggered,
3945 the workstation attempts to authenticate users to afsd_service.exe
3946 by contacting the Domain Controller instead of using the
3947 LSA loopback authentication mechanism.
3949 One of the reasons this bug occurs is because the workstation
3950 does not have a reliable way of knowing that the service whose
3951 netbios name is "AFS" is located on the workstation. This will
3952 be fixed starting in Longhorn Beta 1 by Microsoft. The
3953 "BackConnectionHostNames" registry value will be used to
3954 indicate that the authentications to that service name should
3955 be performed using the loopback authentication mechanism.
3957 In the meantime, when Logon Caching is enabled, we can force
3958 afsd_service.exe to authenticate using the logon cache before
3959 contacting the Domain Controller. This will work with both
3960 password and smart card based logons.
3962 * The allDown logic in cm_ConnByMServers() was wrong. The allDown
3963 flag should not be cleared if a volume's server reference is
3964 marked as "offline". In the case where all of the volume's
3965 servers are either "down" or the volumes are "offline", we want
3966 cm_Analyze() to process the condition as CM_ERROR_ALLOFFLINE
3967 instead of as CM_ERROR_TIMEDOUT. In fact, CM_ERROR_TIMEDOUT
3968 should never occur in practice.
3970 In the case of CM_ERROR_ALLOFFLINE, cm_Analyze() will sleep for
3971 5 seconds, clear the server down and volume busy flags, and
3972 then force an update from the VLDB. This allows the client to
3973 update the location of a volume if the reason for it being
3974 marked offline is because it is being moved. Calls to
3975 cm_ConnByMServers() will be retried either until success or
3976 the RDRtimeout period is reached.
3978 * Correct the Power Management code behavior in response to
3979 Standby, Suspend, and Shutdown events. Instead of flushing the
3980 buffers associated with the mounted SMB submounts, simply write
3981 all buffers which are marked dirty.
3983 * Added support for "TheseCells" to afslogon.dll. "TheseCells"
3984 provides a list of cells other than the default cell for which
3985 tokens should be obtained using the default Kerberos principal.
3986 This functionality is logon domain specific and is only available
3987 when using KFW for authentication.
3989 * Fixed FindNext Invalid Handle error caused by over agressive
3990 attempts at garbage collecting dirSearch entries when the
3991 dirSearch ID wraps from the maximum value to 1.
3993 * Add support for registry defined server preferences for VLDB and
3994 File servers. See registry.txt for details.
3996 * Increased default cache size to 96MB and default number of cache
3999 * Fixed refCount leaks related to directory lookups and pioctl calls.
4001 * Callbacks revoked during a race condition with an attempt to
4002 obtain the same callback no longer result in an inaccessible
4005 * New command line tool:
4007 afsdacl : Set or reset the DACL to allow starting or stopping
4008 the afsd service by any ordinary user.
4010 Usage : afsdacl [-set | -reset] [-show]
4011 -set : Sets the DACL
4012 -reset : Reset the DACL
4013 -show : Show current DACL (SDSF)
4015 * IP addresses are no longer queried once at startup. Instead IP
4016 addresses are obtained as needed. Loopback adapter addresses are
4017 no longer published to the server.
4019 * Pay attention to the MIT KFW registry configuration for automatic
4020 importation of MSLSA credentials. (SOFTWARE\MIT\Leash32,MsLsaImport)
4022 * Fix (once again) case-sensitive comparisons which was apparently
4025 * Activate support for vos listvol -format
4027 * Population of the "AFS Client Admins" group failed on non-English
4028 versions of Windows because the name of the "Administrators" group
4029 is localized. Now we lookup the name of the group by using the
4030 Administrators Group Alias Relative ID.
4032 * Multiple cell token acquisition within afscreds.exe was broken in the
4033 case of Kerberos 5 cross realm authentication. Instead of contacting
4034 the KDC associated with the cell's realm, afscreds would obtain a
4035 token from the KDC of the user's realm. This would result in
4036 "invalid kvno errors" while authenticating to the AFS servers.
4037 Unauthenticated access would work. The symptoms would vary based upon
4038 whether or not the VLDB servers had been contacted using unauthenticated
4039 connections prior to the user obtaining tokens.
4041 * The list of ACL entries was becoming corrupted. It appears as if
4042 an ACL when it expires was not being appropriately cleaned up. In
4043 fact, it was left in the list of ACLs associated with the scache
4046 * Changed the default @sys name list to "x86_win32 i386_w2k i386_nt40"
4047 for 32-bit x86 systems. The default for itanium will be "ia64_win64"
4048 and "amd64_win64" for amd 64-bit processors.
4050 * When executing executables, Windows provides a mechanisms by which
4051 the normal search for DLLs can be bypassed. If a file foo.exe is
4052 being executed and there exists a file or directory "foo.exe.local"
4053 (the contents are ignored if a file), then Windows will search for
4054 DLLs first in the "foo.exe.local" directory and second in the
4055 directory in which the "foo.exe" file is located.
4057 Previous releases of OAFW would improperly return
4058 STATUS_NOT_A_DIRECTORY instead of the expected
4059 STATUS_OBJECT_PATH_NOT_FOUND. This would cause the Windows SMB client
4060 to terminate the search for the DLL causing the execution of the
4061 application to fail.
4063 * Rework the reference counting of smb_vc_t objects. The references
4064 stored in the waiting locks were not counted. This could result in
4065 an assertion if the reference count drops to zero.
4067 * "fs wscell" returns "Freelance.Local.Root" for the workstation cell
4068 when running in freelance mode.
4070 * Added support for persistent caching of file contents, stat entries,
4071 volume data, cell data, and name lookup data. When the registry
4072 value, "NonPersistentCaching", is not defined or set to 0 the file
4073 specified by "CachePath" becomes a persistent cache file. The
4074 size of the cache file is computed at run time. It has a maximum
4075 size of approximately 1.2GB. The contents of the file will be
4076 validated according to the rules specified by the "ValidateCache"
4079 * Added support for UUIDs. UUIDs are kept for as long as the
4080 cache file is intact.
4082 * Added cmdebug.exe and support cache manager debugging callback
4083 interfaces in afsd_service.exe.
4085 Usage: cmdebug -servers <server machine> [-port <IP port>] [-long]
4086 [-addrs] [-cache] [-help]
4087 Where: -long print all info
4088 -addrs print only host interfaces
4089 -cache print only cache configuration
4091 * Symlinks to \\AFS\[all\]... will now be treated the same as
4092 symlinks to /afs/.... However, please use /afs/... as otherwise
4093 the symlinks won't work on Unix.
4095 * Correct a problem with local CellID allocation for cells whose
4096 server list is obtained via DNS instead of a CellServDB file. If
4097 the DNS information expires the CellID assigned to the entry will
4098 be changed. This causes all of the cm_scache_t objects which refer
4099 to the old cellID number to become useless. Attempts to access
4100 files or directories with cm_scache_t objects using the old cellID
4101 will fail since the server list cannot be obtained.
4103 * Correct deadlock condition in cm_EndCallbackGrantingCall which can
4104 be triggered if the volume referenced by the cm_scache_t object is
4107 * The AFS Service needs to respond to SERVICE_ACCEPT_SHUTDOWN messages
4108 in addition to SERVICE_ACCEPT_STOP.
4110 * Move RPC shutdown until after the SMB and RX shutdown procedures
4111 complete. Block until RPC shutdown is complete.
4113 * Modify afslogon.dll (integrated logon) to wait for service if its
4114 state is START_PENDING. If the timeout period occurs, reset to
4115 the retry interval and not the sleep interval.
4117 * When renewing the server list for a cell obtained via DNS AFSDB
4118 records, the cm_cell_t entry must be removed from the list of all
4119 cells. Otherwise, the list of cells will be corrupted.
4121 * In the dcache and scache modules, use the cm_scache_t dataVersion
4122 instead of the cm_fakeDirVersion.
4124 * Synchronize fs.c with the unix version.
4126 * The variable used to determine whether a file or virtual memory
4127 mapped cache is used was not properly initialized to a default
4128 value. If the registry setting "NonPersistentCaching" was not
4129 set, the choice would be random. Properly initialized to be
4132 * The memory mapped view was never unmapped before closing the file
4133 at service shutdown. This is now properly cleaned up.
4135 * Default location of Cache file is now %TEMP%\AFSCache
4139 * A bug affecting new installations of 1.3.75/76 would result in
4140 the creating of incorrect mountpoints in the freelance root.afs
4141 volume for the default cell.
4143 If "fs \\afs\all\<cellname>" lists a volume name of "root" instead
4144 of "root.cell", you have been affected by the bug. To correct the
4145 problem, execute the following commands:
4147 fs rmmount \\afs\all\<cellname>
4148 fs rmmount \\afs\all\.<cellname>
4150 fs mkmount \\afs\all\<cellname> root.cell <cellname>
4151 fs mkmount \\afs\all\.<cellname> root.cell <cellname> -rw
4155 * A bug has been fixed in the auto-generation of Freelance root.afs
4156 symlinks which produced random entries in the \\AFS\all directory.
4158 * Support has been added for multi-homed servers
4161 * Added a new registry value, "StoreAnsiFilenames", which can be used
4162 to force the use of ANSI character sets instead of OEM Code Pages.
4163 This feature is useful when users require the ability to create
4164 filenames with 8-bit characters and need to access the files from
4165 both Latin-1 based Unix systems as well as from Windows.
4167 Activation of this feature will prevent access to files stored with
4168 8-bit OEM characters.
4170 * Shutdown all SMB threads in a synchronized manner when stopping the
4173 * There is currently a maximum cache size of 1.3GB. The limit is imposed
4174 by the largest contiguous block of unused memory within the 2GB process
4175 space which can be assigned to the memory mapped file. Unfortunately,
4176 when the executable digital signature verification code is activated
4177 Windows sees fit to further segment the process memory which in turn
4178 reduces the size of the maximum cache file to less then 800MB. If
4179 larger cache sizes are desired, a new registry value should be set:
4181 HKLM\SOFTWARE\OpenAFS\Client (DWORD) "VerifyServiceSignature" = 0x0
4183 Setting this value will disable the runtime verification of digital
4184 signatures on afsd_service.exe and the afs dlls which it loads. It
4185 will not disable the the version number check on those same files.
4186 The signature verification is not a security messure and is only meant
4187 to enhance the ability to afsd_service.exe to detect potential
4188 destablizing mixtures of DLLs from incompatible distributions.
4190 Added code to auto-disable the signature verification check if
4191 the desired cache size is greater then 700MB.
4193 * Windows' WinTrustVerify(WIN_SPUB_ACTION_PUBLISHED_SOFTWARE) is
4194 used to verify the validity of the afsd_service.exe binary
4195 as well as each of the AFS DLLs loaded by the service. Not only
4196 must the digital signature be valid but the signatures of the
4197 DLL must be signed by the same entity as the service.
4199 * Implement new functions: cm_freelanceMountPointExists and
4200 cm_freelanceSymlinkExists. Use them along with other validity
4201 checks in cm_freelanceAddMount and cm_freelanceAddSymlink to
4202 ensure that name collisions do not occur and that empty strings
4203 are not valid file names.
4205 A symlink may not have a name which would resolve to a valid
4206 cell name. Doing so would prevent access to the cell.
4208 * Add missing cm_HoldSCacheNoLock call to Freelance mount point
4209 re-initialization code. The reference counts of the fake root.afs
4210 volume scache object(s) would become invalid when the mount point
4211 or symlink lists were altered.
4213 * Add registry entries to provide mappings from the afsdsbmt.ini
4214 to the new locations for applications which count on the use
4215 of the old Profile file APIs. These apps are likely to fail
4216 if the user does not have administrator privileges and the
4217 registry is locked down.
4219 * The afs_config.exe submounts dialog had two errors.
4220 First, attempts to remove entries failed because the registry
4221 key was being opened without KEY_WRITE privileges.
4222 Second, when editing a submount entry, changing the name
4223 would add a new key and leave the original one in place.
4224 Now the original submount will be removed if its name is
4227 * In recent months there have been several incidents in which
4228 users have experienced problems starting or accessing
4229 afsd_service.exe and after significant effort has been spent
4230 it has turned out that they have two versions of AFS on the
4231 machine or an inconsistent set of DLLs.
4233 Code has now been added to afsd_service.exe which will walk
4234 the list of modules loaded by afsd_service.exe and validate
4235 that the version of the AFS DLLs matches the version of the
4236 afsd_service.exe executable. If they do not match the service
4239 * When Freelance mode is enabled and there is no registry
4240 key HKLM\SOFTWARE\OpenAFS\Client\Freelance, afsd_service.exe
4241 will attempt to import the afs_freelance.ini file contents.
4242 If the file does not exist, it was creating a dummy file
4243 with a r/o and r/w entry for the default cell and then
4244 importing those values.
4246 This process has been changed. The temporary file is no
4247 longer created. Also, both the OpenAFS Client install
4248 directory as well as %WINDIR% are checked for previous
4249 afs_freelance.ini files.
4251 * Added support for VL_GetEntryByNameN(). Still need to add
4252 support for VL_GetEntryByNameU() for multi-homed support.
4254 * Fix a deadlock situation in afscreds.exe when canceling an
4255 auto-generated Obtain Tokens dialog
4258 * Fix the locking of objects during Directory Searches in the
4259 SMB/CIFS server. The failure to properly lock the reference
4260 counts was resulting in the premature freeing of smb_dirSearch_t
4261 objects while they were still in use by the SMB/CIFS client.
4262 This does not solve the "Invalid Handle" problem.
4264 * Fix Find Cell By Name pioctl call to return a valid cell
4265 name for the Freelance fake root.afs volume.
4266 "Freelance.Local.Root".
4268 * Fix the Explorer Shell Extension Symlinks->Add operation.
4269 The dialog template was missing and the link destination
4270 string was too short.
4272 * Add support for symlinks to Freelance root.afs volume
4273 Stored at HKLM\SOFTWARE\OpenAFS\Client\Freelance\Symlinks
4274 <number> = "<linkname>:<relative-path>."
4275 Use symlink.exe to create, list, or remove
4277 * Remove the fallback to the use of KFW's KRB4 library when obtaining
4278 tokens. We never obtain KRB4 tickets.
4280 * Fix AFS Client Configuration Control Panel to support new SysName
4283 * Fix a bug in afsd_service.exe which could result in the SysName
4284 not being read from the registry.
4287 * Add code to block the issuance of AFS tokens by aklog.exe or
4288 afscreds.exe when the Kerberos 5 principal name contains a dot.
4290 * Modify the IsAdmin() function to always treat the local SYSTEM
4291 account as an AFS client administrator. Affects fs.exe and
4294 * Modify the internal handling of Quota Exceeded errors
4296 * Upgrade all reference count fields in the Windows cache manager
4297 and the osi library to use unsigned long instead of signed short.
4298 A similar fix has been applied to the afs rpc (rx) library.
4300 * fix the Windows cache manager to prevent it from replacing the
4301 rx_connection object associated with the cm_conn_t object on each
4302 and every operation if "fs crypt" was set. This explains the
4303 dramatic performance difference when crypt is used vs clear.
4304 The problem: 'cryptall', a boolean flag indicating whether or not
4305 "fs crypt" is set, was being compared to the rx_connection
4306 cryptlevel which is either rxkad_clear:0 or rxkad_crypt:2.
4307 1 != 2 and therefore the rx_connection was always destroyed
4308 and replaced on each and every operation.
4310 Lock the cm_conn_t object around every call to RXAFS_xxxx functions.
4311 It is not safe for the cm_conn_t object to not be locked because
4312 rx_DestroyConnection might be called from another thread if:
4313 - the user's tokens have changed or expired
4314 - the crypt mode has changed
4316 This fix appears to have also taken care of the problems associated
4317 with Overlapped Writes resulting in Delayed Write errors.
4319 * fix NSIS installer's AdminGroup.exe to properly create and
4320 remove groups when given -create or -remove. The string comparison
4323 * fs sysname now accepts a list of sysname values
4325 * added a new registry value HKLM\SOFTWARE\OpenAFS\Client "IoctlDebug"
4326 DWORD which when set to a non-zero value will cause error message
4327 text to be output to stderr from the pioctl() routine. Useful in
4328 debugging failures of fs.exe, tokens.exe, etc.
4330 * added a test to the power management code to only perform a
4331 flush operation if there is at least one network adapter which
4332 is not a loopback adapter.
4334 * Fix bug in loading of registry value HKLM\SOFTWARE\OpenAFS\Client
4335 "EnableKFW". This value will not be read if the key
4336 HKCU\SOFTWARE\OpenAFS\Client exists; even if the "EnableKFW"
4337 value under that key does not.
4339 * provide mechanisms to force the use of krb524d for Kerberos 5
4340 ticket to AFS token conversion. For afslogon.dll and afscreds.exe
4341 there is a new registry value "Use524" and for aklog.exe a new
4342 command line parameter "-m".
4344 * Fix the pattern matching algorithm to properly match patterns
4347 * smb_ReceiveCoreRename() was factored to produce smb_Rename()
4348 which is used by both the original function and the new
4349 smb_ReceiveNTRename(). smb_ReceiveNTRename() supports the
4350 creation of HardLinks in addition to Renaming. smb_Link()
4351 is a new function which creates HardLinks via cm_Link().
4352 cm_Link() is a new vnodeops function which creates links
4355 smb_ReceiveNTRename() does not support the File Copy and
4356 Move Cluster Information operations described in its interface.
4357 ReceiveNTRename is under documented in CIFS-TR-1p00_FINAL.pdf.
4359 * When opening files via symlinks, we should follow the symlinks
4360 until we reach the actual file stat cache entry. The stat cache
4361 entry of the file should then be stored in the FID instead of
4362 stat scache entry of the symlink.
4364 * return bad operation errors for all unimplemented functions
4365 even if we do not know the functions exist.
4367 * Log bad packets and unknown operation packets to the trace log
4369 * Map CM_ERROR_BADOP to STATUS_NOT_SUPPORTED instead of
4372 * Update list of known CIFS operations to include all those listed
4373 in CIFS-TR-1p00_FINAL.pdf.
4375 * Modify the handling of HKLM\SOFTWARE\OpenAFS\Client\Submounts
4376 to support the REG_EXPAND_SZ type.
4379 * A new Windows authorization group "AFS Client Admins" is now
4380 created and populated with the members of the "Administrators"
4381 group. The group is used to determine which accounts on the
4382 machine may be used to modify the AFS Client Configuration via
4383 the UI and command line tools. afs_config.exe, fs.exe,
4385 * Modify the WinLogon Logoff Event Handler to query NT4 domain
4386 controllers for the remote profile path if Active Directory
4387 services are not available.
4389 * Fix aklog.exe to not add the AFS ID to the username
4391 * PTS registration of new users to foreign cells has been added to
4394 * The cm_Daemon thread is used to perform checks for
4395 down servers, up servers, volumes, callback expirations,
4396 lock maintenance and token expiration. Due to a gaff in
4397 larger integer division the thread never performed any
4398 work. Instead the current time computation would always
4399 be less then the trigger times. This had an adverse affect
4400 on the client's ability to maintain communication with servers,
4401 keep volumes up to date, and flush user tokens and acls
4402 when they have expired. This was broken when the 1.3 branch
4403 was modified to support VC7 which no longer included
4406 * An initialization problem with the Freelance code was
4407 detected while fixing the callbackRequest. The cm_rootSCachep
4408 object is obtained during afsd_InitDaemons() but the callback
4409 information is incomplete. The callback information will not
4410 be obtained until cm_MergeStatus is called from within
4411 cm_GetCallback. Unfortunately, cm_SyncOp did not properly
4412 test for the conditions under which the callback information
4415 * Reports have been filed indicating that callbacks were
4416 being lost. An examination of the code indicated that the
4417 cm_server_t objects were not being properly reference
4418 counted by the cm_scache_t and cm_callbackRequest_t objects.
4419 In particular, the cm_server_t objects may have been freed
4420 from beneath the cm_conn_t objects.
4422 All of the reference counting is now done via the functions:
4427 this improves the ability to track the referrals.
4429 Each cm_BeginCallbackGranting Call now allocates a reference
4430 to the cm_server_t. The cm_EndCallbackGrantingCall either
4431 frees the reference or transfers it to the cm_scache_t
4432 cbServerp field. These are then appropriately tracked
4433 through the cm_Analyze call.
4435 * Ensure that the dnlc hash table is the same size as the
4436 dir name hash table (as per original author's note).
4437 Increase the dnlc CM_AFSNCNAMESIZE to a multiple of 8
4438 for compatibility with 64-bit systems.
4440 * fix smb_ApplyV3DirListPatches to properly apply the hidden
4441 attribute to dotfiles when the infoLevel < 0x101 and
4442 cm_SyncOp has failed.
4444 * Fix the Freelance registry initialization code. There
4445 was a possibility that some systems could end up with
4446 garbage in the registry during a clean install.
4449 * file and directory names beginning with "." will now be given the
4450 hidden attribute when the volume access is anonymous. this matches
4451 the behavior when the volume access is via an authenticated user.
4453 * Added a change monitor to the HKLM\SOFTWARE\OpenAFS\Client\Freelance
4454 key. When a change occurs mark the root.afs data as invalid and
4455 for it to be reloaded on the next access. This allows administrators
4456 to modify the mount point list without restarting the service.
4458 The freelance client used to provide a fake modification time for
4459 the root.afs volume data and its mount points of 7/09/2001 14:24 EDT.
4460 Added code to extract the last modification time of the Freelance
4461 registry key and use that instead. The time now represents the
4462 most recent mount point change.
4464 * PTS registration of new users to foreign cells has been added to
4467 * Additional Cache Control and Credential Manager options have been
4468 added to the WiX installer. See deployment guide for details.
4470 * The CachePath setting is now optionally a REG_EXPAND_SZ type
4472 * The WiX installer has been upgraded. Version 2.0.1927.1 is now
4475 * The loopback installation code may have had a problem updating the
4476 %ETC%\HOSTS file which could have resulted in a premature failure.
4477 Work around code has been added for the case where the file cannot
4480 * The default max chunksize was increased from 15 (32K) to 17 (128K)
4481 because Windows sends 64K blocks when using overlapped writes.
4483 * The default number of server threads was increased from 4 to 25 to
4484 better handle overlapped writes.
4486 * The "AfscredsShortcutParams" registry value was not being properly
4487 loaded by afscreds.exe. Therefore, the default value was always being
4488 used instead of the value set by the installer.
4490 * Windows XP provides downgrade attack detection to prevent an attacker
4491 from being able to force the use of NTLM simply by disrupting
4492 communication with the KDC. This attack cannot exist between the
4493 Windows CIFS client and the AFS Client Service. Therefore, when a
4494 downgrade has been detected the afs pioctl library will force the
4495 establishment of a new CIFS connection using NTLM.
4497 * A locking error was discovered surrounding all references to volume
4498 server lists within the cm_cell.c source file.
4500 * The logged into Windows username was incorrect on Terminal Server
4503 * A new registry value "NonPersistentCaching" was added to the service
4504 parameters key. When set to a non-zero value, the afs cache is stored
4505 in the Windows paging file. There are two limitations to choosing
4507 1. when persistent caching is implemented it won't work with
4508 this flag set since there will be nothing to persist.
4509 2. with this flag set the initial paging allocation cannot be
4510 changed while the service is running
4512 * An initialization bug was discovered in aklog.exe which affected users
4513 who have a domain name for their afs servers which could not be mapped
4517 * afs_config.exe now validates cell names against DNS in addition
4518 to the CellServDB file.
4520 * In order to allow the freelance client to connect to a volume with ID
4521 equal to 1 on the default cell we changed the fake root.afs volume ID
4522 once again. This time we choose 0xFFFFFFFF. In addition, we change
4523 the cell ID of the fake root.afs volume from 1 to 0xFFFFFFFF as well.
4524 It will now be impossible for a volume ID to match that of another
4525 cell unless the client is connected to 0xFFFFFFFD cells. That should
4526 be enough room for growth.
4528 * Fix "fs mkmount" command to work with UNC paths and when
4529 started from non-AFS drives. It is now possible to create a mount
4530 point in the freelance fake root.afs volume with the command
4532 fs mkmount \\AFS\all\<directory-name> <volume-name> <cellname>
4536 fs mkmount \\AFS\all\openafs.org root.cell openafs.org
4537 fs mkmount \\AFS\all\.openafs.org root.cell openafs.org -rw
4539 * The algorithm used to re-attempt access to the servers associated with
4540 a volume has been altered to properly address the case in which all
4541 servers have been marked down. The previous algorithm did not reset
4542 the server's down flags so the servers were never actually retried.
4543 This caused a problem with active volumes if the network connectivity
4544 was lost as could be the case with a network cable removal, wireless
4545 drop, or laptop hibernation. With the fix volume access is restored
4546 almost instantenously when network connectivity becomes available.
4548 * Support for SMB/CIFS browsing has been added to the AFS Client Service
4549 SMB server. It is now possible to use "NET VIEW \\AFS" to obtain a
4550 listing of AFS submounts and freelance mount points. Support for
4551 NETSHAREENUM, NETSHAREGETINFO, NETSERVERENUM2, NETSERVERGETINFO
4552 significantly enhances the behavior of AFS volumes within the Explorer
4553 Shell. For instance, "AFS" now shows up as server in the Explorer
4554 with each submount or freelance mount point visible as a share.
4555 The right click menu in each folder now works with full functionality
4556 on a consistent basis.
4558 * The network provider can be configured to have different behavior
4559 depending on the domain that the user logs into. These settings are
4560 only relevant when using integrated login. A domain refers to an
4561 Active Directory (AD) domain, a trusted Kerberos (non-AD) realm or the
4562 local machine (i.e. local account logins). The domain name that is
4563 used for selecting the domain would be the domain that is passed into
4564 the NPLogonNotify function of the network provider. (see registry.txt
4567 * Added a new registry value [HKCU\SOFTWARE\OpenAFS\Client]
4568 "Authentication Cell" which may be used to specify a default
4569 authentication cell for afscreds.exe which is different from
4570 the default cell for the AFS Client Service daemon.
4572 * Added a Logoff WinLogon Event Notification function to afslogon.dll.
4573 afslogon.dll moved to %WINDIR%\System32\.
4574 New registry entries added to register the dll for Winlogon events.
4576 The logoff event will now force a call to ktc_ForgetAllTokens()
4577 using the context of the user being logged off as long as the
4578 user's profile is not loaded from within AFS. If the profile
4579 was loaded from AFS we can't release the tokens since the Logoff
4580 event is triggered prior to the profile being written back to
4581 the its source location. This is now performed in an XP SP2
4584 * Windows XP SP2 Internet Connection Firewall interoperability
4587 * The %WINDIR%\afsdsbmt.ini contains four sections:
4588 Submounts, Drive Mappings, Active Maps and CSC Policies.
4589 The Submounts and CSC policies are now stored in the registry under
4590 [HKLM\SOFTWARE\OpenAFS\Client\Submounts]
4591 [HKLM\SOFTWARE\OpenAFS\Client\CSCPolicy]
4592 The Drive Mappings and Active Maps are stored in the registry under
4593 [HKCU\SOFTWARE\OpenAFS\Client\Mappings]
4594 [HKCU\SOFTWARE\OpenAFS\Client\Active Maps]
4596 There is no automatic migration of this data as it would be impossible
4597 to consistently migrate data to user profiles which may not be active
4598 when the machine is updated.
4600 * The %WINDIR%\afs_freelance.ini contains lists of mountpoints for the
4601 fake root.afs volume. For the same reasons as for the cellservdb file,
4602 this information should not be in %WINDIR%. This information is now
4603 kept under the registry key
4604 [HKLM\SOFTWARE\OpenAFS\Client\Freelance]
4606 The data from the afs_freelance.ini file will be automatically
4607 migrated to the registry on first execution of afsd_service.exe
4609 * Keeping the CellServDB file in the location %WINDIR%\afsdcell.ini is
4610 troublesome for several reasons. One, it is confusing for those who
4611 expect the file to be named "CellServDB" instead of "afsdcell.ini".
4612 Two, this file is not a Windows Profile formatted file. Three,
4613 applications should not be reading or writing to %WINDIR%. It causes
4614 problems for Windows Terminal Server.
4616 The new location of CellServDB will be the OpenAFS Client install
4617 directory which is by default C:\Program Files\OpenAFS\Client and can
4618 be determined by querying the registry for
4619 [HKLM\SOFTWARE\TransarcCorporation\AFS Client\CurrentVersion]PathName
4621 The existing afsdcell.ini will be migrated by the NSIS installer.
4622 The Wix installer must still be updated to do the same.
4624 * Change NSIS installer to use DNS by default; to remove Integrated Logon
4625 High Security mode; and to add Terminal Services compatibility registry
4626 entries to allow the OpenAFS tools to find the afsdcell.ini and other
4627 configuration files in %WINDIR%.
4629 * Add support for authenticated SMB connections. This will remove
4630 the need for high security mode in most situations. Both NTLM
4631 and Extended Security (GSS SPNEGO) modes are supported. Effectively,
4632 only NTLM can be used even though Kerberos is now supported. The
4633 reason is that it is not possible to construct a service principal
4634 which is unique to each individual machine.
4636 SMB Extended Auth does not work on XP SP2 unless one of two registry
4637 modifications are made:
4639 (1) To disable the check for matching host names on loopback connections
4640 set this key. This does not require a reboot:
4642 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa]
4643 "DisableLoopbackCheck"=dword:00000001
4645 (2) To add the AFS SMB/CIFS service name to an approved list. This
4646 does require a reboot:
4648 [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa\MSV1_0]
4649 "BackConnectionHostNames"=multi-sz "AFS" "MACHINE-AFS"
4651 afsd_service.exe will automatically add the current Netbios Name
4652 to the BackConnectionHostNames list and then temporarily disable
4653 the loopback check for one cycle of startup/shutdown of the service.
4654 We assume most folks do not start/stop without a reboot so this
4655 will be adequate in most cases.
4657 * Fix security hole in afslogon.dll which allowed passwords to be
4658 sent in clear text to the KDC in a misformed principal name.
4660 * Fix cm_GetCell() to properly handle expired dns entries
4663 * If Freelance mode is active and the afs_freelance.ini
4664 file does not exist, do not create an empty file.
4665 Instead create a file containing ro and rw mountpoints
4666 to the default cell using the standard conventions.
4668 * Modify the Freelance support to handle the ability
4669 to create rw mount points in the fake root.afs volume.
4671 * Changed the RPC mechanism used for token setting from
4672 named pipes to local. Use of named pipes can be restored
4673 by setting the environment variable AFS_RPC_PROTSEQ to
4676 Named pipes were required when a Windows 9x system was
4677 using a NT system in gateway mode which is incompatible
4678 with our use of local loopback adapters.
4680 * In afscreds.exe, if a username of the form user@REALM is
4681 specified and no password is specified, do not perform a
4682 kinit operation. Only perform the aklog functionality.
4684 * Add a new registry value which allows the number of processors
4685 on which afsd_service.exe executes to be restricted. Valid
4686 values are 1..numOfProcessors
4688 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
4692 * A second MSI based installer option is now available.
4694 * Fixed Kerberos 5 kinit functionality in afscreds.exe to properly
4695 request tickets for user/instance@REALM instead of just user@REALM
4697 * Modify the Power Management Notify routine to wait for the Hard Dead
4698 timeout period instead of a fixed 19 seconds. With the longer timeout
4699 periods Hibernation and Standby could never succeed when network
4700 connectivity is not available.
4702 * The following fs.exe commands are now restricted to Administrator:
4703 - checkservers with a non-zero timer value
4706 - sysname with a new sysname list
4715 setting the default sysname for a machine should be done via the
4716 registry and not via "fs sysname".
4718 * NSIS installer adds options to install Debugging Symbols
4719 and the Microsoft Loopback Adapter; the user is now also
4720 given the ability to select the afscreds.exe startup options.
4722 * Build system modified to generate symbols for FREE (aka RELEASE)
4723 builds as well as CHECKED (aka DEBUG) builds
4725 * Sites which have a volume ID of 0x20000001 assigned to their
4726 root.cell volumes have been experiencing problems with accessing
4727 the root.cell volume of their cell when Freelance mode has been
4728 active. This was because 0x20000001 was assigned to the fake
4729 root.afs volume created by freelance. The fake volume id is
4730 now set to 0x00000001 to prevent conflicts.
4732 * The timeout logic in the AFS Client Service has been wrong
4733 for sometime. It is based on two different assumptions.
4734 First, the SMB client timeout is a fix value as was the case
4735 with OS/2 Lan Manager. This assumption is incorrect. The
4736 SMB timeout in Windows is a dynamic value computed based upon
4737 a fixed minimum timeout to which is added time based upon the
4738 size of the request and the performance characteristics of
4739 the connection. Second, it is the responsibility of the
4740 SMB Server to enforce the timeout requirements of the client.
4741 This is untrue. The SMB Server cannot be expected to know
4742 the requirements of the client. More importantly, if the
4743 SMB server uses the SMB client timeout as a value to restrict
4744 its behavior as an RX client, the performance characteristics
4745 of the local SMB session would be used to prematurely terminate
4746 WAN connections with significantly different performance
4749 The timeout logic has therefore been modified in the following
4751 . the Lan Manager Workstation (SMB) Session Timeout is used only
4752 as a basis for configuring the Connection Dead Timeout
4753 and Hard Dead Timeout values. The Connection Dead Timeout
4754 must be at least 15 seconds longer than the SMB Timeout
4755 and the Hard Dead Timeout must be at least double the
4756 Connection Dead Timeout.
4757 . New registry entries have been added to allow the Connection
4758 Dead Timeout and Hard Dead Timeout values independent of the
4759 Lan Manager Workstation Session Timeout
4760 . The test to enforce the SMB Client Timeout has been removed.
4762 One of the side-effects of removing the enforcement of the SMB
4763 Client Timeout is that regardless of whether or not the SMB client
4764 is available to receive the response (and how would the SMB server
4765 know) the RX protocol response can be used to update the AFS
4766 Client Service state for ready access by future SMB client
4769 This should be the end of the "Server paused or restarting messages"
4771 * Add support for arbitrary UNC paths to the pioctl() support.
4772 This enables the fs commands as well as the AFS Shell Extension
4773 to work correctly when UNC paths are being used.
4775 * Fix afscreds.exe (by updating afskfw.lib) to search for cells via
4776 DNS if the cell configuration cannot be determined via CellServDB
4778 * Add debug info to test whether CM_BUF_WAITING or CM_SCACHE_WAITING
4779 are ever set more than once at a time
4781 * Fix the management of lists of cm_cell_t structures when using
4782 DNS to lookup cell information. The previous code would fail to
4783 reuse the same cellID for a cell if DNS was used more than once
4784 for a given cell name. When the ttl expired, a single cm_cell_t
4785 could be inserted into the cm_allCellsp list more than once
4786 producing a loop. In addition, the vlServerp list belonging to
4787 the cell was not freed resulting in improper refCounting of the
4790 * Add DNS support to cm_IoctlNewCell() which previous only examined
4793 * Add cm_FreeServer() function and call it from cm_FreeServerList()
4794 to properly garbage collect cm_server_t objects
4796 * Add numVCs variable to smb.c to track the number of smb_vc_t
4797 objects created and use it to initialize the vcID field which
4798 previously was set to 0 in all objects resulting in FindByID
4801 * Fixed DNS lookups to work consistently throughout the OpenAFS
4802 product instead of just from within the afsd_service.exe
4804 * Added a runtime check to ensure that AFS Client Service SMB
4805 Server is accepting connections before attempting to mount
4808 * Read IP addresses for volume servers out of the CellServDB
4809 file if gethostbyname() on the hostname fails.
4811 * Fix getcellconfig() to populate both the Hostnames as well
4812 as IP addresses when loading cell data via DNS
4814 * Increase the Connection Dead Time to 50 from 20 seconds
4815 Increase the Hard Dead Time to 120 from 40 seconds
4816 (matches the Unix values)
4818 * Fixed an assertion validating the number of allocated NCBs
4820 * Fixed the build environment to consistently build for
4821 Windows 2000 and above. (APPVER = 5.0)
4823 * Fixed rx_debug to properly validate the receipt in incoming
4824 data with select() and recvfrom(). Do not copy data out of
4825 the socket buffer unless success is indicated.
4828 * afsd_service.exe will now display a message box to the
4829 desktop when it terminates due to an IP Address Change.
4831 * installer no longer deletes AFS Server configuration data
4834 * installer generates a warning dialog if the RPC service
4835 is not properly configured
4837 * installer compressed with lzma instead of bzip2
4839 * afsd_service.exe shutdown crash solved once and for all
4841 * reference counting of smb_vc_t data structures improved
4843 * name space collision of smb_fid_t event objects corrected
4845 * the output of "fs memdump" is now written to
4846 %WINDIR%\TEMP\afsd_alloc.log
4848 * the file TaAfsApp_1033.dll is now properly installed allowing
4849 the User Manager to start
4851 * a new algorithm is used for computing filename pattern matches
4853 * afscreds.exe now accepts user names containing instance
4856 * Fix the Directory Name Lookup Cache to be case-sensitive.
4857 This is crucial in environments in which a Windows client
4858 is accessing a directory with more than one filename that
4859 differs only by case. If the directory contains "FOO"
4860 and "Foo". You want "DEL Foo" to delete the correct one.
4861 We still have a problem in that "DEL foo" will delete a
4862 random filename. This will be addressed in a future release.
4864 * Fix afscreds.exe -M option (renewMaps) to work when High
4865 Security mode is off. Also, remember to disable the ActiveMap
4866 flag in afsdsbmt.ini when a drive mapping is removed.
4868 * Updates to NSIS installer script. AFS Server configuration
4869 data will not be destroyed on un-install or re-install.
4870 Use a better compression algorithm.
4872 * afslogon.dll now uses KFW to obtain tokens when available
4874 * afslogon.dll when given an all uppercase username will
4875 attempt to authenticate with both the uppercase name
4876 and an all lowercase variation
4878 * DST modification removed. The fix appears to make things
4879 worse after a reboot of the machine.
4881 * fs.exe: added "cscpolicy" which is used to
4882 change client side caching policy for AFS shares
4884 Usage: fs cscpolicy [-share <AFS share>] [-manual] [-programs]
4885 [-documents] [-disable] [-help]
4887 * Several uninitialized variables have been initialized
4889 * It is now possible to obtain tokens using cross realm
4890 Kerberos within afscreds.exe:
4892 user: jaltman@ATHENA.MIT.EDU
4894 Will obtain a cross realm ticket for jaltman/DEMENTIA.ORG@ATHENA.MIT.EDU
4895 will will in turn be used to obtain afs@DEMENTIA.ORG.
4896 The resulting token will be stored with the display name
4897 jaltman@ATHENA.MIT.EDU@dementia.org
4899 * aklog.exe has been added to the client
4901 Usage: aklog [-d] [[-cell | -c] cell [-k krb_realm]]
4902 [[-p | -path] pathname]
4906 -d gives debugging information.
4907 krb_realm is the kerberos realm of a cell.
4908 pathname is the name of a directory to which you wish to authenticate.
4909 -noprdb means don't try to determine AFS ID.
4910 -5 or -4 selects whether to use Kerberos V or Kerberos IV.
4911 (default is Kerberos V)
4912 No commandline arguments means authenticate to the local cell.
4915 * All of the resource files have been restructured to adhere to
4916 a set of rules IBM implemented for loading string resources.
4917 These rules had either been forgotten or were not discovered
4918 by folks working on the OpenAFS sources. The end result was
4919 memory corruption. This is primary item which was preventing
4920 the AFS Server from working.
4922 * Increased the size of the maximum ticket size stored in a token
4923 from 344 bytes to 12,000. Increased the buffers used to convey
4924 messages between the pioctl() caller and the SMB Server from
4925 1000 bytes to 12,512. The code appeared to have been writing
4926 above the top of the stack by quite a few number of bytes.
4927 (The increased ticket size is necessary for the next item.)
4929 * When obtaining AFS Tokens via KFW, krb524 is no longer required.
4930 Instead the raw Kerberos 5 ticket is used in its entirety. This
4931 is extremely important as it allows us to use pure Kerberos 5 KDCs
4932 as the source of the AFS authentication. The use of up to 12,000 byte
4933 tickets will allow tickets produced by all versions of Microsoft
4934 Active Directory to be used.
4935 - create a user account.
4936 - designate it DES only
4938 - specify its UPN to be "afs@realm"
4939 - assign a SPN of "afs/cellname" to the UPN with setspn.exe
4941 * Do not enforce the funky 8dot3 pattern matching rule that the first "."
4942 is special when using long file names. (you must use "*.*" and not "*")
4943 Instead only enforce it when performing 8dot3 searches.
4945 * Fixed the DST problem with creation times being set one hour ahead
4947 * Fixed the problem when using \\afs\cell-alias. For example,
4948 \\afs\uncc instead of \\afs\uncc.edu. Do not a new cell struct
4949 for the alias name; instead simply expand the name. One of the
4950 symptoms of this problem was a loss of acquired tokens.
4952 * Fixed the AFS Shell Extension. The Symbolic Link menu was empty
4953 of strings. (Only English strings provided.)
4955 * Fixed the installer to properly replace in use files.
4957 * Fixed the build system to cleanup generated component version files
4959 * The release build compiled with MSVC 6.0 compiler to avoid the
4960 afsd_service.exe shutdown crash. This does not solve the problem
4961 but simply avoids it for the time being.
4965 * fix afslogon.dll to not corrupt memory when High Security mode
4968 * fix afsd_service.exe to not attempt to restore the stack when
4969 an exception occurs. (not safe in multi-threaded programs)
4971 * fix uninstaller to properly remove the CRT and MFC DLLs
4973 * remove a Message Box from afscreds.exe when getcellconfig()
4974 fails on a kerberos realm which is not a cell
4976 The following is a list of changes to the OpenAFS for Window client
4979 * "fs setserverprefs" will leave afsd service deadlocked
4981 * "vos listaddrs" will core dump
4983 * installer sets the appropriate keys to support Integrated Logon
4985 * installer disables the "Find Lana by Name" functionality as it
4986 was causing headaches for many users
4988 * fix the intermittent crash of the power management thread when
4989 shutting down the AFS Client Service
4991 * optimizes the obtain drive mount list functionality which is
4992 executed every time the mount tab in afscreds.exe and afs_config.exe
4993 are refreshed. (this happens a lot)
4995 * fix the service shutdown logic. add the STOP_PENDING state
4996 and do not accept additional service events after we declare
4999 The following is a list of changes to the OpenAFS for Window client
5002 * flexelint was run against the source tree and hundreds (perhaps
5003 thousands) of corrections were applied to ensure prototypes
5004 were in use; types were used consistently; variables were
5005 initialized; unused variables were removed; etc.
5007 * A wide variety of instrumentation was added including the
5008 ability to produce a stack trace from within afsd_service.exe
5011 * Dynamic configuration of the RDRtimeout value based upon the
5012 LanMan Workstation Session Timeout
5014 * The mount root no longer needs to be called "/afs". This
5015 is now set by a registry value "MountRoot" within the key
5016 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5018 * The cell list is now only read out of afsdcell.ini when the
5019 file changes instead of each time a cell is resolved.
5021 * Thread synchronization was added to cm_server.c and ktc_nt.c
5023 * All calls to GlobalAlloc()/GlobalFree() were replaced with
5024 calloc()/free(). The Global functions were needed on Windows 3.x
5025 but have caused a variety of problems on the Win32 platforms.
5026 Avoiding them is highly recommended by several Microsoft
5027 Knowledgebase articles
5029 * Support for Symbolic Links added to the AFS Shell Extension
5031 * Added a registry value "OverlayEnabled" to determine if
5032 Shell Extension Overlays should be enabled.
5033 HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
5035 * New Build system to support VC6, VC.NET, VC.NET2003 compilers and
5036 separate trees for checked and free builds. Build system supports
5037 a custom directory src\WINNT\extra which can be used as a grafting
5038 location of organization specific additions to the build tree.
5040 * New installer built using NSIS 2.0.
5042 * Named all kernel objects in order to allow them to be monitored
5043 with tools such as SysInternals' ProcExp.exe.
5045 * Introduced new EventLog framework for AFSD
5047 * Introduced Power Management interface to AFSD for Standby and
5048 Hibernate modes to allow cache to be flushed prior to network
5051 * Utilize Win32 DNSQuery API instead of internal routines. This
5052 allows DNS SRV queries to be sent to all current domain name
5053 servers. Not just one specified in an INI file. DNS is now
5056 * "NetbiosName" registry value may be used to specify a fixed
5057 Netbios Name such as "AFS" to be used instead of "HOSTNAME-AFS"
5058 when the loopback adapter is in use. If you need to use the
5059 old notation with a loopback adapter installed specify a registry
5062 "NetbiosName" REG_EXPAND_SZ = "%COMPUTERNAME%-AFS"
5064 * Refactor all modules which depend on LAN Adapter and NetbiosName
5065 determination in a new library: lanahelper.lib. This allows for
5066 consistent behavior throughout the product.
5068 * Move the afsd.log and afsd_init.log files to the directory specified
5069 by the "TEMP" environment variable. This is usually %WINDIR%\TEMP
5070 for services. Added the Date to the log entries.
5072 * New registry value "RxMaxMTU" used to limit the size of the RX
5073 packets sent by the AFS Client Service to the Server. In order
5074 to enable OpenAFS to work across the Cisco IPSec VPN the packet
5075 size must be restricted to 1264 or smaller. The latest NSIS
5076 installer sets a value of 1260 by default.
5078 * New registry value "RxNoJumbo" to disable the use of Jumbo Rx
5079 packets. This is not needed in order to work across the Cisco
5080 VPN but might be needed for other network environments. This
5081 value is not set by the NSIS installer.
5083 * New registry value "HideDotFiles" is used to apply the Hidden
5084 attribute to files whose names begin with a '.'. This value
5085 is set by the NSIS installer.
5087 * New registry value "MaxMpxRequests" allows the maximum number
5088 of multiplexed sessions to be configured at run time. This
5089 value is not set by the NSIS installer. The default value is
5092 * New registry value "MaxVCPerServer" allows the maxmimum number
5093 of VCs per server to be configured at run time. This value is
5094 not set by the NSIS installer. The default value is 100.
5096 * New registry value "AllSubmount" allows the "all" submount to
5097 be disabled by setting its value to 0x00.
5099 * Allow cells names to be valid mount points
5100 \\<netbiosName>\<cellname>
5102 * Store the active state of drive mappings in order for afscreds.exe
5103 to restore them upon startup
5105 * Add exception handling to generate a Stack Trace to the afsd_init.log
5106 file if one happens to occur.
5108 * Add lots of logging to help detect the cause of invalid SMB packets
5110 * Enable Kerberos for Windows to be used to obtain AFS Tokens via
5111 conversion of Kerberos 5 "afs" service tickets. Supports auto-
5112 renewal of expiring tokens as long as afscreds.exe is running.
5114 * New afscreds.exe command line options:
5116 -M = renew drive maps
5117 -N = ip address change detection
5120 * New registry value "EnableKFW" in {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5121 determines whether or not MIT Kerberos for Windows should be used
5122 to obtain tokens via Kerberos 5 tickets.
5124 * New registry value "AfscredsShortcutParams" in
5125 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5126 determines the command line parameters to be specified when "fixing"
5127 the AFS Shortcut in the user's startup folder.
5129 * The "ShowTrayIcon" registry value has been moved from
5130 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5131 {HKCU,HKLM}SOFTWARE\OpenAFS\Client
5133 * The <cell name> registry values used to store the token expiration
5134 reminders have been moved from
5135 HKLM\Software\TransarcCorporation\AFS Client\AfsCreds to
5136 {HKCU,HKLM}SOFTWARE\OpenAFS\Client\Reminders
5138 * Obtain the Logon User Name from the Explorer key when available
5140 * new text document doc\txt\winnotes\registry.txt lists all registry
5141 values used by OpenAFS (excluding the AFS Server)
5143 * BUG: rx_securityClass objects were not properly reference
5144 counted and were never freed.
5146 * BUG: reduce the number of conditions under which CM_ERROR_TIMEOUT
5147 would be generated. The existence of a server does not imply
5148 that it is not down. If all of the servers for a cell are down
5149 return CM_ERROR_NOSUCHVOLUME instead. This prevents the Explorer
5152 * BUG: the directory name lookup cache failed to free the entries
5153 in the cache when the name cache entries cycled. The entries
5154 in the cache would become dereferenced without being freed.
5156 * BUG: fs setserverprefs could be executed without Administrator
5159 * BUG: the number of allocated NCB objects (100) exceeded the number
5160 which could actually be waited upon by the kernel (64). Any objects
5161 which were utilized above the limit could never have event completions
5164 * BUG: smb_username_t objects were not being reference counted and
5165 were not properly freed.
5167 * BUG: smb_tid_t objects could under unusual circumstances be freed
5168 before they were no longer referenced.
5170 * BUG: smb_fid_t object pointer were frequently used even when
5171 their value could be NULL. They were not properly released and
5172 therefore they were never freed.
5174 * BUG: smb_packet_t data structures were not completely initialized
5177 * BUG: when Rx produces a CM_ERROR_NOIPC error do not return "Access
5178 Denied" because that causes the Explorer Shell to try again until
5179 access is obtained. Instead return "Remote Resources" which allows
5180 the shell to move on and treat the error as transient.
5182 * BUG: when initializing the NCBreturns structure, separate Event objects
5183 were created for each NCB although a single Event object was supposed
5184 to be shared by all.
5186 * BUG: smb_dirSearch_t objects were not being properly referenced counted
5189 * BUG: smb_tran2Packet_t objects were not being properly referenced
5192 * BUG: directory path creation did not handle the case of multiple
5193 directories requiring creation in one attempt
5195 * BUG: SMB requests which required an Extended Response were ignored.
5196 This prevented some files from being written to AFS volumes.
5198 * BUG: character strings were being freed even after they were
5199 inserted into in use data structures
5201 * BUG: inconsistent usernames were used when High Security mode was
5202 enabled. (there is still much to do in this area)
5204 * BUG: pioctl() calls which require out of band RPC operations were
5205 susceptible to race conditions when performed by multiple processes
5207 * BUG: memory allocation and deallocation crossed instances of the
5208 C Runtime Library producing memory leakage and corruption in
5209 afscreds and the client configurator.