[openafs.git] / doc / txt / winnotes / afs-changes-since-1.2.txt

Since 1.6.0002 (1.6.0b)

 * Correct an error in mount point string construction
   when the user provides a cell alias to "fs mkmount"
   instead of the full cell name.  Every mount point
   string must end in a trailing dot.

 * Major improvements to the Explorer Shell Extension.

 * Addition of the AFS Redirector drivers and service
   interface.

Since 1.6.0001 (1.6.0a)

 * Correct an error in CheckOfflineVolumeState processing

 * Permit NPLogonNotify() to exit immediately if the
   SetToken error is KTC_NOCM.

Since 1.6.0000 (1.6.0)

 * Avoid a cm_serverLock reference count leak.

Since 1.5.9907 (1.6pre7)

 * Add Shutdown message to the windows event log

 * Check Offline Volume status according to the specified policy
   not every time the daemon thread runs.

 * if the directory object cannot be found in the tree, return
   CM_ERROR_PATHNOTFOUND instead of crashing.

 * avoid memoryt leak of cm_GetSpace() allocations

 * improve afs/kfw error message output.  Include PT and KTC
   errors.  Impacts afslogon.dll, afscreds.exe and afssrvadm.exe

 * afslogon.dll NPNotifyLogon will now kick start the
   afsd_service if it is not already in a pending start or
   started state.

 * properly construct the userrealm in the afskfw library.
   do not include the '@' in the realm.

 * fix the generation of HTML reference manual pages from
   POD.

 * No longer send release lock RPCs to a file server for
   a deleted file.  Each attempt will fail with a VNOVNODE
   abort.  Multiple aborts can result in the file server
   throttling the client.

 * cm_BkgDaemon should not perform the job of cm_SyncOp
   which already properly serializes operations on the
   a file.

 * cm_BkgDaemon should not perform queued tasks on deleted
   objects. Doing so is a waste of resources.

 * Avoid rx nat pings prior to connection attachment.

 * After directory enumeration completes move the directory
   status object to the most recently used position to avoid
   it being recycled when the directory contains more objects
   that the status cache can hold.

 * Reduce memory fragmentation caused by lock order validation
   checks.

 * When a status object is deleted, move its object to the
   least recently used position in the queue to promote
   recycling when the reference count drops to zero.

 * Fix the osi_Log macros so they safe for use in if..else
   statements by wrapping in do..while(0).  Several logged
   messages were wrong as a result and there may have been
   other logic errors caused by mismatched if and else
   statements when braces were not used in the code.

 * Apply interlocked operations to all state and queue fields
   in the cm_call, cm_volume, cm_scache, and cm_buf objects
   in order to ensure that operations are atomic.  State flag
   bits were being lost on multi-processor systems under heavy
   load.

Since 1.5.9906 (1.6pre6)
 * add support for detection of NTFS symlinks to
   \\AFS using GetFileInformationByHandleEx which is
   available on Vista, Server 2008 and beyond.

 * improve thread safety of afs_shl_ext.dll

 * MS11-043 adds response validation for SMB_COM_NEGOTIATE
   messages received by the SMB Redirector.  OpenAFS failed
   to properly specify a Challenge and DomainName in the
   response when the security mode is SMB_AUTH_NONE (or share
   with password).  This patchset corrects smb_ReceiveNegotiate()
   so that it adheres to the protocol specification.

 * smb_T2SearchDirSingle() must not fail directory search requests
   for the _._AFS_IOCTL_._ file.  Although this file does not actually
   exist, it is successfully processed by CreateFile operations.
   Therefore, an explicit search for it should return a valid answer.

 * cm_SyncOp/cm_SyncOpDone is used to synchronize the RPC processing
   to ensure that calls which are in conflict cannot occur at the
   same time but also to ensure that the ordering of operations
   is consistent.  cm_MergeStatus() was in many cases executed after
   cm_SyncOpDone() removed the synchronization barrier which in turn
   permitted status information to be applied out of order.  Side
   effects could have included data loss due to client side file
   truncation.  More commonly two StoreData RPCs would have their
   status information applied out of order forcing the cache manager
   to invalidate all of the cached data for the file.

 * cm_NewServer() can result in a call to cm_UpdateVolumeLocation()
   if a server probe is performed.   In order to avoid recursive
   calls to cm_UpdateVolumeLocation() do not probe new servers from
   within cm_UpdateVolumeLocation().

 * Numerous Rx improvements.

 * Fix out of order lock acquisition in smb_ReceiveV3LockingX()
   which occurs if the lock acquisition fails.

Since 1.5.9905 (1.6pre5)
 * aklog supports dotted Kerberos v5 principal names.

 * afskfw library always attempts afs/cell@USER-REALM

 * afskfw library must test return code from
   krb5_cc_start_seq_get() or will trigger a null
   pointer exception when using Heimdal.

 * lock protected fields must be 32-bit in order
   to avoid memory overwrite races.

Since 1.5.9904 (1.6pre4)
 * Fix caching of non-existent volumes.  The test to
   trigger an immediate CM_ERROR_NOSUCHVOLUME in
   cm_UpdateVolumeLocation() was backwards.

 * Prevent the background daemon from checking the
   status of non-existent volumes.  cm_CheckOfflineVolumes()
   should skip volume groups with the CM_VOLUMEFLAG_NOEXIST
   flag set.

 * The afskfw library should return an error immediately
   if the krb5_32.dll library cannot be loaded.  Affects
   afslogon.dll and afscreds.exe.

 * No longer depend on leashw32.dll in afskfw library.

 * NPLogonNotify must provide the user password in all
   calls to KFW_AFS_get_cred().  It cannot count on a
   credential cache being preserved between calls.  Permits
   tokens to be acquired for all cells listed in the
   TheseCells registry value for a domain.

 * Improve the trace logging from NPLogonNotify().

 * Avoid a race when writing the cm_scache_t mountPointString
   when acquiring mount point or symlink target data via
   cm_GetData().  The race could result in bogus target
   data being cached.

 * Permit the use of des-cbc-md5 and des-cbc-md4 enctypes
   as DES keys in asetkey.exe.

Since 1.5.9903 (1.6pre3)
 * Fix automatic addition of Freelance cell mount points
   broken in 1.5.9902.

 * Avoid recursive offline volume checks which could cause
   a crash due to stack exhaustion.

Since 1.5.9902 (1.6pre2)
 * rx calls issued on a busy call channel are now specially
   handled and logged.  All calls are automatically retried.

 * The vos support for multi-homed servers was incorrectly
   implemented.  The changes have been reverted.

 * Avoid wasting cm_buf_t objects and associated data buffers
   for obtaining mountpoint and symlink target strings.  Instead
   read them directly into the cm_scache_t mountPointTarget
   buffer.

 * Optimize the evaluation of mountpoint and symlink targets
   by skipping the FetchStatus and a full round-trip if
   the cache manager does not have valid status information
   for the object.

 * Fix the abstraction of cm_FreelanceAddMount() by passing
   in volume names that do not include the trailing dot.

Since 1.5.78
 [there was no 1.6pre1 for Windows]

 * vos commands now manipulate servers by UUID and can
   recognize multi-homed servers.

 * afs_config will not longer set the Tray Icon State
   in the registry if the checkbox is not present in
   the dialog. [RT 128591]

 * Heimdal's roken utility library has been added
   as \Program Files\Common\afsroken.dll

 * When probing servers to determine if they are up or down
   no longer issue an RXAFS_GetTime RPC to servers that indicated
   that they do not understand the RXAFS_GetCapabilities RPC.
   Since they responded it is known that they are up.

 * AFS Explorer Shell Extension now works from folder
   backgrounds.  Overlays for mount points and symlinks
   are present in the dll, but are not registered at present
   by the installers.

 * Do not use RankServerInterval registry value as the value for
   PerformanceTuningInterval.

 * CellServDB updated to 13 Dec 2010 release from grand.central.org

 * Add "fs chmod" command and display current mode as part
   of "fs examine" output.

 * When the data version of a mountpoint or symlink changes,
   the target string in the cm_scache_t object must be cleared.
   Otherwise, the new target will not be queried.

 * "fs checkservers" now includes vldb servers in the output
   and only lists multi-homed servers once.  A multi-homed
   server that has at least one up interface is no longer
   considered to be down.

 * When asynchronously storing dirty data buffers to the
   file server ensure that (a) the cm_scache_t object and
   the cm_buf_t object are for the same File ID so that
   locking and signalling work properly; and (b) if the
   FID no longer exists on the file server, do not panic,
   just discard the buffer.

 * When processing VNOVOL, VMOVED and VOFFLINE errors perform
   server comparisons by UUID or address and not simply by
   cm_server_t pointer.  Otherwise, server failover may not
   succeed.

 * Do not preserve status information for cm_scache_t objects
   when the issuing server is multi-homed.

 * Giving up all callbacks when shutting down or suspending
   the machine is now significantly faster due to the use
   of an rx_multi implementation.  (This functionality is
   still off by default and must be activated by a registry
   value.)

 * Race conditions were possible when updating the state
   of the cm_volume_t flags and when moving the volumes
   within the least recently used list.

 * Ensure that the lanahelper library does not perform a
   NCBRESET of each lan adapter when enumerating the
   current network bindings.  Correcting this permits OpenAFS
   to work on Windows 7 when the network adapter settings
   change.

 * Fix creation of mount points and symlinks as \\AFS\xxxx


Since 1.5.77
 * Reimplement FetchData and StoreData operations to make
   use of rx_Writev and rx_Readv which is faster than than
   multiple calls to rx_Write and rx_Read.  Throughput
   improvements of up to 20% are possible.

 * Many improvements to the rx rpc stack:

   - RTT measurements

   - Fast recovery is not triggered by packet resend timeouts

   - Socket errors are now detected and reported

   - avoid repetitive memory allocations

   - avoid calling gettimeofday() for every packet

   - reduced dependency on the call->lock

   - When a call receive is completed, ack all of the packets
     immediately.  Do not wait for the first response packet
     to be sent.

   - Do not hold the call->lock during rx_Write* and rx_Read*
     operations when the call fields in use can only be accessed
     by the application thread.

   - Reduce the default window size from 128 back to 32 because
     the current overhead from walking packet queues in the rx
     listener thread significantly hampers the ability to
     process incoming packets from the network.

   - Only backoff the peer timeout once for a series of naks.

   - Release builds are lean and mean.  All debugging and
     packet tracking code is disabled.

 * Version number is logged to the Windows Application Event
   Log as part of the start pending message.

 * MSI installers now backup and restore configuration
   data during upgrades.

 * Rx pmtu discovery has been disabled by default.  Use
     HKLM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
       RxPMTUDiscovery "DWORD"
   value to re-activate.

 * cm_BkgDaemon event evaluation has been optimized to avoid
   unnecessary overhead.

 * If a volume is known to be readonly, then locally fail any
   change request without sending it to the file server.  This
   avoids triggering the file server's abort threshold.

 * Do not leak cm_volume_t objects from the volume LRU queue.

 * Ensure that cm_NameI errors are acted upon promptly.
   There are many cases in the SMB server where an error from cm_NameI()
   was either ignored or not acted upon until several other operations
   are performed that could result in the same error being repeated.
   This is a mistake which did not have negative side effects until
   additional checks for callback status were added recently.

   At present, if a CM_ERROR_ACCESS error is returned and ignored,
   subsequent attempts to operate on the same cm_scache_t will result
   in additional queries to the file server that will also end in an
   abort response.  This can trigger the file server to delay responses
   to the client.

 * Fix fs mount point and symlink create operations which were
   failing due to improper directory separator termination.

 * RXAFS_GetVolumeStatus can return VNOVOL, VMOVED, etc.  In
   order to process them and update volume state a fid must
   be passed to cm_Analyze().  Use the volume root fid.

 * Negative caching for volume lookups that fail with either
   VL_NOENT or VL_BADNAME.  The expiration time is five minutes.
   This prevents volume lookup storms.

 * Provide each volume lookup its own two hour lifetime instead
   of reseting all volume location information every two hours.
   This prevents the first smb path evaluation after the volume
   location data has been discarded from potentially causing an
   smb client timeout.

Since 1.5.76
 * When the SYSTEM account is in use, the SMB connection is
   not always established as S-1-5-18.  Sometimes it uses an
   anonymous, S-1-5-7, connection.  In both cases the SMB
   authentication name is the nul string.  To better identify
   when an SMB connection is local system, the smb_username_t
   is now stored as a SID instead of a user name.  When a SID
   is used, a new flag bit is set SMB_USERNAMEFLAG_SID.

   In smb_SetToken if AFS_PIOCTL_AFSLOGON is set and the user
   connection is not SYSTEM, the RPC SID is checked.  If RPC
   SID is SYSTEM, that supercedes the SMB authentication.

 * Replace STATUS_IO_TIMEOUT with RPC_NT_SERVER_TOO_BUSY for
   CM_ERROR_ALLBUSY.

 * Replace STATUS_PATH_NOT_FOUND with RPC_NT_SERVER_UNAVAILABLE
   for CM_ERROR_ALLOFFLINE and CM_ERROR_ALLBUSY.

 * Optimize behavior of buf_CleanAsyncLocked().  Avoid searching
   for the cm_scache_t object by FID if it is already known.

 * If the readonly file attribute is set (stored as a unix mode)
   then a CreateFile operation should fail if the file is opened
   for DELETE in combination with any other privilege.

 * If the directory buffer contents are garbage we can crash
   the service.  Add some simple validation checks to ensure
   that cm_dirEntry_t objects have the correct flag value and
   that the name strings are not too long.

 * If the bulkStat errorCode indicates that a particular object
   is inaccessible due to a VIO error, we must update the server
   status appropriately in order to permit failover.

 * An RX_MSGSIZE error is returned by the new PMTU detection
   code.  It is critical that such an error result in a retry of
   the operation that failed.  Otherwise, the PMTU detection can't
   work and the server will be marked down.
   Secondly, it is important that such errors not leak to the
   application layer.  Map them to CM_ERROR_RETRY in all cases.

 * Translate RX_RESTARTING to the same cache manager error code
   for all RPC classes.

 * When logging server volume instance errors to the windows
   application event log, be sure to log the cell as well.
   Translating from server ip address is non-trivial.  Make it
   easier for administrators triaging issues to plug the volume
   and cell info into vos commands.

 * Since unix mode bits are represented in octal in most cases
   make sure we log them that way.

 * The cm_enforceTrailingDot() function failed if there was a
   trailing dot followed by white space.

 * Rx: Do not send RX_MSGSIZE if the PMTU size did not change.

Since 1.5.75
 * A crash could occur if a single letter share name was
   dynamically evaluated as matching a cell name.  This was
   due to a failure to treat the comparison of two nul
   strings as identical.

 * Docbook validation is performed by xmllint during builds.

 * Volume package bugs in the file server can result in VOFFLINE
   being returned to the client instead of VNOVOL or VMOVED.  As
   a result the Unix CM treats VOFFLINE the same as VMOVED and VNOVOL.
   The Windows client has not.  As a result, bugs in the file server
   can cause the Windows client to lose if the volume has in fact
   been moved to another server.

   As part of this change, the volume location list is updated prior
   to the volume status being applied to the server from which the
   error was received.

 * If the pages to be flushed are from a readonly or backup volume
   they can't be dirty.  In this case there is no need to stabilize
   the pages before they are locked and recycled.
   Stablilization is performed on the cm_scache_t object so do not
   stabilize and unstablize for each cm_buf_t object.

 * Over the years the processing of the Freelance callbacks have
   added functionality that behaves much more like FetchStatus checks
   to a file server.  If the data version of the object has changed,
   get the new data.  Given that is the case, we can remove much of
   the original refresh logic that is rather race prone.   Say goodbye
   to cm_fakeGettingCallback and cm_fakeDirCallback.

 * When processing a pioctl path with either smb_ParseIoctlPath or
   smb_ParseIoctlParent, cm_SyncOp(CM_SCACHESYNC_NEEDCALLBACK|GETSTATUS)
   must be called on the cm_scache_t object to ensure that it is up
   to date before we permit cm_Lookup or other operations to be performed
   on it.  Add the cm_SyncOp() call to smb_ParseIoctlPath and
   smb_ParseIoctlParent to ensure it is done for all pioctl operations.

 * The 32-bit tools installer was displaying client configuration wizard
   pages.  Since, the 32-bit tools share configuration with the AFS
   service, the 32-bit tools installer should not be attempting to change
   client configuration.  Remove said wizard pages from the 32-bit tools
   installer.

 * cm_data.rootSCachep is a global pointer to the cm_scache_t that
   represents the root.afs volume root directory.  Throughout the
   code this cm_scache_t was being used without ensuring that a
   callback to the volume is in fact valid.

   Under most circumstances this would not be a problem.  However,
   it is possible for a request to fail due to the lack of a callback
   at a critical moment.  Add a new function cm_rootScachep() that
   attempts to ensure that a callback is present (if possible) prior
   to use the cm_scache_t object.

 * When the contents of the Freelance root directory changes the fake
   directory buffers are updated and a fakeDirVersion is incremented.
   The dataVersion of the cm_scache_t object is supposed to be updated
   on the next access by performing a fake get callback request.
   Unfortunately, this did not always occur because of a race.  If another
   Freelance object is updated first, the root directory object would
   never successfully get a fake callback.

   This patchset ensures that the generation of the fake directory
   buffer content and the callback are obtained under the same set
   of locks thereby removing the race.

 * Enforce in afsd_service that tokens can be set for alternate
   userids only if the SetToken pioctl is called from the
   Local System account.

Since 1.5.74
 * Revise SMB QuerySecurityInfo Response for MS10-020

   MS10-020 (http://support.microsoft.com/kb/980232) has caused
   many problems for implementors of SMB 1.0 servers and applications
   that call GetFileSecurity() without checking the return code to
   determine if the call succeeded.  The gist of the vulnerability
   was that the SMB redirector would pass any buffer it received
   to the application regardless of whether or not it was valid.
   MS10-020 protects the applications by strictly validating the
   SMB response data structure and the data in the security descriptor
   that is returned.

   The problem for SMB 1.0 server implementors is that there have
   been at least three different protocol descriptions for
   NT_TRANSACT_QUERY_SECURITY_DESC published over the last decade
   and all of them are incomplete.  Therefore, just about no one but
   Microsoft has an SMB 1.0 server implementation that produces the
   exact out that they are expecting to validate.

   The end result is that in an attempt to protect applications from
   crashing due to invalid input being passed in directly caused
   dozens of applications to crash by not returning any security
   descriptor data at all.  Even when the applications didn't crash
   they might not have been able to save their data.  Cisco WAAS
   and NetApp DataOnTap systems were most adversely affected and
   they have had CIFS protocol licenses for many many years.

   To fix OpenAFS here is what needed to be done:

   1. Instead of returning a security descriptor that gives ownership
      to the NUL SID, give it to the Everyone SID and set the flag
      that states that everyone has full access.

   2. Validate the input parameters.  In particular, check to ensure
      that the SMB file descriptor is valid and the file has not
      been deleted.

   3. Enforce the maximum output data and parameter counts.

   4. Handle buffer overflow and buffertoosmall conditions
      in the manner that Microsoft expects them to be handled.
      In particular, note that the parameter data which is returned
      in the SMB Data Region is not counted in the Data Count.
      Even if MaxData is 0, we can still return parameters values
      as long as MaxParm is large enough.

 * Prevent use of AFSCache file contents if mapped to
   a new address.

 * The Windows version of "fs newcell" did not accept any parameters
   and behaved quite differently from the Unix version.  Instead of
   permitting new cell information to be added, the Windows version
   simply forced the existing cell information to be reacquired.

   This update adds a new pioctl, VIOCNEWCELL2, to support the
   implementation of a Unix-style "fs newcell".  The functionality
   added here differs from the Unix version in the following ways:

     1. "fs newcell" with no arguments is still accepted
        in order to maintain compatibility with prior Windows
        behavior.

     2. "fs newcell -cell <cell> -dns" instructs the cache manager
        to add the new cell but obtain the vldb server info from
        DNS.

     3. "fs newcell -cell <cell> ... -registry" instructs the cache
        manager to add the new cell and also save the cell configuration
        data in the registry for use the next time the service restarts.

     4. The -vlport and -fsport options are accepted although the
        -fsport value is currently unsupported by the cache manager.

 * New registry value "FreelanceImportCellServDB" instructs Freelance
   to create a mount point for every cell name listed within the
   CellServDB.

 * Path MTU discovery for Rx is activated.

 * Rx socket input buffer is converted to a circular buffer.

 * Fix usage of cm_FreeServerList().  Do not set the server list
   pointer to NULL after calling cm_FreeServerList().  Doing so
   can result in a memory leak.

 * Only enable Rx NAT pings on a single anonymous connection at a
   time.

 * Fix cm_IoctlSkipQueryOptions() buffer management.  Prevents a
   potential read beyond end of memory buffer.

 * Reduce requested privileges when reading registry CellServDB
   to the minimum required.

 * Add support for RPC Pipe Service NetWkstaGetInfo levels
   101 and 102 which are called on Windows 7 and 2008-R2.

 * Prevent integer overflow during quota percent used calculation
   in Explorer Shell Extension  (RT 126846)

 * Generate a meaningful error if "fs listacls" or "fs setacls"
   are executed on the Freelance root.afs volume.

 * RXAFS_InlineBulkStat errors must be processed via cm_Analyze.
   RXAFS_InlineBulkStatus does not return errors such as EACCES,
   VNOVOL, VNOVNODE, VOFFLINE, VBUSY, VIO, VMOVED, etc. as an RPC return
   code.  Instead they are returned in the status info errorCode field
   for each file.

   Traditionally, the error associated with the first FID in the query
   list has been returned to the caller of cm_TryBulkStatRPC().
   However, the error has never been processed through cm_Analyze()
   which means that the per-vnode processing for VNOVNODE and the volume
   global processing for VMOVED, VNOVOL, etc. has never been performed.
   As a result, failover to other .readonly volume instances cannot occur,
   volume moves will not be handled, and files that have been deleted
   are not detected.

   This patchset makes the following changes:

   1. If an inline bulk operation has been performed and the inline
      errorCode is a volume global error, then that error replaces
      the RPC return code within the cm_Analyze() processing for
      the RPC.  This will affect whether or not a retry operation
      is performed.

   2. The variable 'inlinebulk' is reset to 0 at the top of the
      cm_Analyze() loop in case failover from an inlinebulk capable
      file to an inlinebulk incapable file server takes place.

   3. The FID that is passed into cm_Analyze() is not a real fid.
      Instead it consists of the cell and volume but vnode = 0.
      This ensures that the error (if any) is not applied to the
      directory object.

   4. If an inline bulk operation was performed, prior to performing
      the cm_MergeStatus() operation a vnode a check is made to
      determine if an error was returned for that vnode.  If so,
      cm_Analyze() is called with no connection, a fake cm_req_t,
      the fid, and the error.  This permits cm_Analyze() processing
      to be performed on the file.  

 * Show configuration pages for all types of MSI installations

   The OpenAFS MSI installer wizard used to not show any configuration
   pages for "Typical" and "Complete" installations.  Setting the
   workstation cell and logon options during installation required
   selecting the "Custom" option.  Many users choose the "Typical" option
   during installation, and thus would never see the configuration pages.
   Therefore, for these users, the workstation cell was being set to the
   default.

   This patch makes the workstation cell and logon option configuration
   pages visible to all types of installations (except silent
   installations which show no UI).

 * cm_LookupInternal creates Freelance mount points and symlinks
   when queries cannot be found in the Freelance root.afs directory.
   If the search name is a full cell name for which vldb information
   can be obtained, then a mount point is added.  If the search name
   is a left-most substring or the full cell name with a dot appended
   to it, then a symlink was created.  This approach created a very
   poluted Freelance name space.

   This patchset makes the following changes:

    1. Do not create symlinks with a dot appended to the cellname

    2. Do not create symlinks where the left-most substring is not
       a full dot separated component of the cellname.

    3. Permit lookups to succeed when we would have created a
       symlink in the past without creating the symlink.

 * BPlus tree lookups are much faster than searching through
   the native directory format on Windows because the case sensitive
   hash tables cannot be used successfully.  Permit BPlus trees
   to be used except when called with cm_BPlusDirFoo as the action
   function because cm_BPlusDirFoo is used to build the BPlus trees
   from the native directory format.

 * Symlinks are ending up in the Freelance root.afs directory that
   end with a dot.  Make sure it cannot happen.

 * cm_FreelanceAddMount and cm_FreelanceAddSymlink is supposed to
   return the allocated FID of the entry that was added.  However,
   cm_NameI is called to perform the lookup without forcing an update
   of the Freelance fake directory.  As a result the entry may not be
   found.

   Force an update prior to calling cm_NameI() by using
   cm_clearLocalMountPointChange() and cm_reInitLocalMountPoints()
   if required.

 * The Freelance fake root directory buffers were not zero-filled.
   This results in random behavior that can cause the service to
   terminate unexpectedly.

 * The validation check for the response from the GetVolumeStatus
   pioctl is incorrect.  The response is not simply a VolumeStatus
   structure but also several C strings appended to it.

 * When flushing a file, we need to commit the file length changes
   as well as the dirty buffers.  Call cm_FSync instead of buf_CleanVnode
   which is called by cm_FSync.

 * Prevent rx_rpc_stats global lock from being a bottleneck in the
   Rx library.

Since 1.5.73

 * Avoid a race when updating cell vldb server lists
   that can result in a crash.

 * Avoid a deadlock when managing CM_SCACHESYNC_STOREDATA
   state operations for directory objects.

 * Add new Windows Application Event log messages for
   VBUSY, VRESTARTING, ALL_BUSY, ALL_OFFLINE, and ALL_DOWN.
   Include message throttling to prevent the same message
   from being logged repeatedly within a five second window.

 * Reduce lock contention by waiting for cm_buf_t I/O
   operations to complete before permitting cm_SetupStoreBIOD
   to analyze a buffer for inclusion in a BIOD.

 * Split the cm_buf_t flags field to separate the flags
   that are protected by the cm_buf_t mutex from those
   protected by the buf_globalLock.  This eliminates the need
   to hold both locks everytime the flags field is accessed.
   Both locks were not held in the past resulting in race
   conditions that could result in deadlocks.

 * Add "vos setaddrs" command.

 * Rx library lock contention avoidance between rx_NewCall and
   rx_EndCall.

 * Rx library races due to inconsistent use of rx_connection
   conn_data_lock to protect the flags field.

 * Rx library inconsistent use of RX_CALL_TQ_WAIT which could
   result in deadlocks.

 * Rx library must signal transmit queue waiters when flushing.
   Otherwise, deadlocks can occur.

 * In cm_UpdateVolumeLocation, avoid searching for a ".readonly"
   volume on a numeric volume name.

 * File buffer allocations whose offsets are beyond server EOF
   should be locally allocated and zero filled.  The file server
   should not be issued a FetchData rpc which is guaranteed to
   fail.

 * Enable integrated logon to work with Windows 7/2008 when
   user logons are performed with a non-Domain Kerberos principal.

 * Add Protection Error messages to aklog output.

Since 1.5.72

 * Prevent the Explorer Shell extension from crashing if
   symlink creation failed.  (126406)

 * A Rx level NAT ping has been implemented.
   Add NatPingInterval registry value to
     HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters
   to permit Nat Ping to be enabled.  The default value is 0 seconds.

 * When a re-initialization is taking place, be sure to reset
   cm_noLocalMountPoints to 0 in case someone deletes the "Freelance"
   registry key out from underneath the service.

 * Add krb5 error message translation to aklog, afscreds,
   afslogon.dll, the network identity manager afs provider
   and translate_et.

 * Mode bits aren't directly exposed by the Win32 API. We were leaving
   them to default to 0777 when creating new files and directories.
   This version introduces two configuration parameters;
   'UnixModeFileDefault' and 'UnixModeDirDefault' which are DWORD
   registry entries that are used to set the initial mode bits.
   If the values are set to 0, then the behavior is identical to what we
   had before.

 * Minidump files are now produced with a timestamp appended
   to the name.

 * An SMB request debugging monitor has been added.  When activated
   the monitor will automatically turn on trace logging if any SMB
   request has required longer than 60 seconds to complete and will
   then create a minidump every 60 seconds thereafter until the
   request completes.

Since 1.5.71

 * Restore use of DNS AFSDB and SRV records by kaserver clients.

Since 1.5.70

 * Avoid a potential Freelance deadlock during initial execution 
   of afsd_service.exe if the old ini file data has to be
   imported.

 * Three rx library corrections.  (1) Idle data connection
   processing could timeout if the send window filled and
   took longer than the idle data timeout period for the
   transmit window to re-open.  (2) The transmit queue
   could be emptied prematurely.  A required check for the
   queue being in use was forgotten.  (3) The function that
   is supposed to implement a wait for the transmit queue
   to cease being busy failed to wait.

Since 1.5.69
 * Restore use of DNS AFSDB and SRV records which were
   unintentionally disabled in 1.5.69

Since 1.5.68
 * Add a context menu to the NetIdMgr AFS Provider 
   notification icon.

 * Prevent an empty directory Btree from being created and
   marked as valid if cm_BPlusDirBuildTree fails.

 * cm_BPlusEnumAlloc should not fail if the enumeration
   contains zero entries.

 * In cm_BPlusXXX functions, return ENOMEM if malloc() fails.

 * Do not leak the cm_scache_t dirlock if cm_BPlusDirBuildTree
   fails.

 * fs examine should report owner and group ids as signed values.

 * Protect buffers in smb_WriteData from simultaneous writes.

 * Query network interface MTU values from the registry because
   the IP Helper GetAdapterAddresses() fails to report a manually
   configured MTU.

 * Improve performance by not dropping and reacquiring the rx_call
   lock in rx_WriteProc* and rx_ReadProc*.

 * When configuring the rx library, the network interface MTU
   must be applied to both the maximum send size as well as the
   maximum receive size as is done when rx_SetMaxMTU is called
   by the application.

There was no 1.5.67 release

Since 1.5.66
 * Convert afsauthent.dll to use the xdr_alloc and xdr_free
   routines from afsrpc.dll in place of malloc and free.
   This permits the two libraries to be built with different
   C runtime libraries without crashing.

 * Set the DOS Readonly attribute on a file/directory 
   whenever the unix mode combined with the mask 0200 
   is true.  Previously there was a discrepency between
   the mask used for testing for readonly behavior and
   that used for setting the attribute.

 * Disable AFSVolSync based .readonly "whole-volume callback"
   support because the all file servers prior to 1.5.67
   (and perhaps 1.4.12) do not properly assign a value to
   the AFSVolSync structure in bulk status RPC responses.

 * Improve the error output from aklog to output the value
   from krb5 error_message() if the afs_com_err output
   indicates an unknown value.

 * Hold a lock on the cm_scache_t object when the smb
   server is assigning a new clientModTime value.

 * Permit custom build version numbers to be set
   without modifying the src/config/NTMakefile.<platform>
   files.

 * Add support for binary code signing with cross-signed
   certificates and arbitrary versions of signtool.exe.

 * Convert VBUSY and VRESTARTING to CM_ERROR_ALLBUSY and
   do not permit them to be exposed to the smb redirector.

 * Convert STATUS_TIMEOUT responses to STATUS_IO_TIMEOUT
   to avoid confusion within the smb redirector.

 * Fix the byte order assigned to port numbers associated
   with AFSDB record lookups.  They must be network byte 
   order not host byte order.

 * Add dynamic server ranking based on RPC round trip
   time measurements.

Since 1.5.65
 * Official support for Windows 7 and Server 2008 R2

 * Improved Rx RTT computation following the lessons
   learned by TCP.

 * Prevent a file server bug (FetchData returning an
   invalid length instead of zero) from causing an
   "unexpected network error" when writing to files.

 * Promote DNS SRV records as superior to DNS AFSDB
   records.  Support arbitrary port numbers for vldb
   servers.

 * Add "fs listacl -cmd" support.

 * Add AFSVolSync based .readonly "whole-volume callback"
   support.  With this functionality, multiple objects from
   a .readonly volume can have their status validated by
   issuing a single RXAFS_FetchStatus RPC.

 * Remove drive mapping functionality and service start/stop
   from afscreds.exe.

 * Remove drive mapping functionality from afs_config.exe.

 * Adjust SMB error return codes to avoid returning 
   STATUS_TIMEOUT which results in the SMB redirector
   disconnecting.

 * Network Identity Manager OpenAFS Provider now provides its
   own "AFS lock" notification icon to report the status of
   "have tokens, have no tokens, service not started, service
   started but inaccessible".  Hovering over the icon lists the
   cells for which tokens exist (if any) and the OpenAFS version
   number.  Double-clicking executes the Network Identity Manager
   default action.

 * When merging the current status for an AFS object, ensure
   that the volume status for that object is set to 'vl_online'.

 * Permit cm_Analyze to request a retry even when the cm_req_t is
   flagged CM_REQ_NORETRY if the RPC failed due to RX_CALL_DEAD.
   The retry will force the use of a new connection.

 * Add digital signatures to resource only dlls as required by
   Windows 7.

 * Prevent pioctl calls from retrying indefinitely when a sharing
   violation error occurs.

 * Add {HKLM,HKCU}\SOFTWARE\OpenAFS\Client  DWORD  "ShowMountTab"
   to restore access to drive mapping functionality in afscreds.exe
   and afs_config.exe.

 * Permit Rx minimum peer timeout, maximum receive window size and
   maximum send window size to be configured via the registry.

 * The MSI installer now sets the ARPINSTALLLOCATION as required by
   Windows 7.

 * DNS lookups set the port numbers in network byte order within
   the cache manager.

 * Replace use of the 16-bit compatibility API WinExec with
   ShellExecuteEx in afs_config.exe and afscreds.exe.  WinExec
   is incompatible with Windows 7 / Server 2008 R2 UAC 
   functionality.

Since 1.5.64
 * Short circuit background volume checks if a shutdown
   or suspend request is received.

 * Avoid contacting the file server in order to allocate
   an empty buffer beyond the current length of the file.

 * When a request for streams on a directory or mount 
   point object is received, do not offer a default stream.
   This was affecting the saving of roaming profiles.

Since 1.5.63
 * Fix Windows 2000 compatibility.  Do not call rand_s() 
   even if it is compatible with the C run time library
   version.

 * Fix a data consistency error between the output of 
   NetWkstaGetInfo and NetServerGetInfo RPCs, specify the Lan 
   workstation group name "AFS", and report server name as 
   "AFS" instead of "\\AFS" when the caller asks for "\\AFS".

 * Enable executables to be run from \\AFS on Windows 7.  
   Return "Name not found" instead of "File not found" when 
   a directory or file name cannot be found.

 * Prevent cache manager from marking file server "down" when 
   the data returned in response to either RXAFS_FetchData64 
   or RXAFS_StoreData64 is invalid.

 * Add pioctl data validation to the AFS Explorer Shell extension.

Since 1.5.62
 * After 1.5.62 afslogon.dll would no longer crash but it 
   also would not load "domain" specific configurations.
   Fixed.

 * An uninitialized variable was present in the symlink
   recursion detection loop.

 * Reverse addition of unique per file GUID during 
   Create/Open responses. Inclusion of the GUID prevents 
   Cygwin applications (and others) from accessing //afs.

 * Treat "filename::$DATA" as a synonym for "filename".  
   This addresses the recent report of roaming profiles 
   failing to save back to the file server.

 * When given a choice between an error from rx_Write() 
   or RXAFS_EndStoreData(), use the rx_Write() error.  
   This ensures that over quota and access denied errors 
   are acted upon.

 * If an error occurs during RXAFS_StoreData operations, 
   apply that error to all of the buffers in the BIOD. 
   This ensures that failures will not be retried if they 
   will only fail again.

 * Do not perform a DNS lookup for a cellname if the 
   cellname prefix is _._AFS_IOCTL_._.  This will avoid 
   DNS lookups when the query is for 
   _._AFS_IOCTL_._.F7E5F580200909010061TTTTNT7TT.{10E39A49-4531-4496-A08E-842D4C440D20}.

 * Fix Freelance root.afs volume object callback processing.  
   Object synchronization and status merging bugs.

 * Fix Explorer Shell "invalid parameter" bug introduced 
   in 1.5.62.

 * Log RXKAD errors to the trace log.

 * Prevent simultaneous pioctl calls from multiple processes 
   from stepping on each other's data.

 * Prevent simultaneous pioctl calls from crashing the 
   afsd_service.

 * Add data validation to pioctl output processing in 
   ktc_GetToken and ktc_ListTokens.

 * Add data validation to pioctl output processing in 
   fs.exe and symlink.exe.

Since 1.5.61
 * IMPORTANT: New support for DCE RPC Services: SRVSVC and WKSSVC.
   No longer will browsing \\AFS produce truncated share names.

 * Registry specified server preferences did not result in
   the CM_SERVERFLAG_PREF_SET flag being applied to the
   cm_server_t object.  

 * Avoid unnecessary DNS lookups of share names as cells
   when it is known that the name cannot be a cell name.
   Any name that does not contain a dot is skipped.

 * When processing Dfs Referral requests, do not return an
   error if the server is down or busy; if the volume is
   offline; the cell vldb cannot be reached; etc.
   These paths are still \\AFS paths and so the client 
   should not be sent to search elsewhere for resolution.

 * CRITICAL: Prevent the cm_Daemon thread from terminating when
   the machine enters the suspend state.  IF the cm_Daemon thread
   dies, there is nothing to execute down server checks.

 * Prevent use of smb_StartedLock before initialization which
   can result in a panic in the lock package.

 * Improve error checking in afslogon.dll in order to prevent
   NULL pointer dereferences within GetLogonDomainOptions()
   if the Lsa operations fail.

 * Add support to the build system to automatically update
   a Microsoft Symbol Store.

 * Do not access the cm_conn_t after executing the cm_Analyze
   loops.  Doing so can result in an access to memory that has
   been freed.

 * CRITICAL: If a StoreData request offset is not aligned on the 
   buffer module blockSize, (offset % blockSize) bytes of dirty
   data will not be written to the file server.

 * CRITICAL: If a StoreData64 request is sent to a file server
   that does not support large files (>2GB) and the client is 
   unaware that the server has no large file support, the first 
   chunksize worth of data will not be written to the file server.
   This bug was introduced in 1.5.3.   It affects all IBM AFS file
   servers and all OpenAFS file servers older than 1.4.0.

Since 1.5.60
 * If a file server becomes inaccessible while the cache manager has 
   dirty buffers to write, the afsd_service buf_IncrSync thread can 
   attempt to use 100% of the cpu.

 * Fix the locking that protects the cm_cell_t lists 
   (all, name, id, free).

 * Remove cm_cell_t objects from the name and id hash tables before 
   placing them on the free list.

 * Fix "fs newcell" which was broken in 1.5.60.

 * Do not attempt to synchronize dirty buffers if the associated 
   volume is known to be unavailable.

 * Log to the Windows Event Log dirty buffers that cannot be written 
   during the shutdown process.

 * Modify behavior of a Freelance mountpoint target that does not 
   specify a cell.  Instead of assuming the target volume is in the 
   Freelance.Local cell, use the workstation "Cell" specified in the 
   registry. A mountpoint target of "#root.cell." will now mean the 
   root.cell volume in the workstation cell for the current session.  
   If the workstation cell changes from "athena.mit.edu" to 
   "andrew.cmu.edu", the referenced volume will also change without 
   requiring that the mount point targets be altered.

 * Avoid false infinite loop errors when validating the cm_cell_t 
   free list during startup.

 * Fix two more locations where xdr_free() should be used instead of 
   free() to permit a checked built of afsd_service.exe to work with 
   release builds of afsrpc.dll.

 * Add cm_FindServerByUuid().  Re-implement RXAFS_InitCallBackState3() 
   to permit the server Uuid to be used to lookup the server object 
   and from that determine the cell.  This permits callbacks that are 
   received from alternate addresses to be processed with a known server 
   object.  Previously a request from an unknown server would clear all 
   callbacks from all cells.

 * Fix a bug that prevented optimal performance when using a non-zero 
   value for 'daemonCheckVolCBInterval'.  As a reminder, when 
   "daemonCheckVolCBInterval" is set to a non-zero value, all .readonly 
   volume callbacks are automatically renewed 90 minutes before their 
   expiration.

 * Fix automatic ranking of vldb servers whose values are obtained from 
   the CellServDB file.

 * Add logging for RX CALL DEAD errors which are generated if the remote 
   host is unreachable.

 * Add failover for RX CALL TIMEOUT errors when the volume is readonly 
   or the call is to a vldb server.

 * Add registry based cell search functionality to NetIdMgr, 
   afs_config.exe, and klog.exe.

 * afsconf_GetCellInfo() has been modified to perform gethostbyname() 
   lookups on the host names in the CellServDB instead of using the 
   specified IP addresses.  This provides aklog, pts, vos, etc. the same 
   CellServDB behavior that the Windows Cache Manager uses.

 * Completely a behind the scenes change.  The snprintf() routines have 
   been completely rewritten.

 * When updating the stat cache entry callback of a .readonly object 
   from the volume group object, update the file server reference to 
   ensure it matches the most update to date callback.

 * Add proper support for processing callbacks from multi-homed file 
   servers.  Instead of comparing servers by cm_server_t pointer, 
   compare them by UUID when the UUID is known.

 * During a shutdown short circuit the offline volume check daemon 
   functionality.

 * Properly track link counts during cm_Link and cm_Unlink operations.

 * Return the error code of RXAFS_FetchData / RXAFS_StoreData in 
   preference to an error code reported by rx_EndCall.

 * Add "PerFileAccessCheck" registry value to permit testing against 
   experimental file servers that include per-file acl support.  This 
   value is intentionally undocumented.  It is not to be used by
   production environment deployments.

 * Log a warning to trace log if a volume group is marked as a DFS 
   File Set.  The Windows OpenAFS client does not include DFS support.

 * Fix a bug introduced in 1.5.60 that prevents the afs netidmgr 
   provider from obtaining tokens when referrals are in play.

 * Add "fs chown" and "fs chgrp" commands to permit the owner and 
   group of objects stored in AFS to be set from Windows.

 * Avoid performing background daemon operations when the machine is 
   going into suspend mode.

 * Perform offline volume checks in most recently used order.

 * Prevent crash when a data version for a cache object goes backwards.

 * fs setquota, fs setcachesize, vos setfields, and vos create now 
   accept human readable orders of magnitude. (K, M, G)

 * fs listquota fixed to permit large quota sizes to be displayed.

 * rx packet resend and data packets sent counts were incorrect.

 * Multi-thread safe library versions are now being generated and used.  
   mtafsubik.lib, mtafsutil.lib, mtafsvldb.lib, mtafsvol.lib.

 * Microsoft SMB Redirector (mrxsmb.sys) support for ExtendedSessTimeout
   values are now available on XP through Windows 7.  Add functionality
   to autodetect if such support is present on the machine.  If so,
   configure it if necessary and dynamically adjust the AFS Rx timeout
   values accordingly.

 * Correct another race condition in the Rx library that could result
   in an unexpected panic while freeing the Rx call iovq.

Since 1.5.59
 * A fix to the pioctl library to support drive substitution 
   to UNC paths.  (SUBST <d:> <\\afs\cell\path>).

 * On April 9th Microsoft released a Hot Fix for Windows Server 2003 SP2 
   that corrects a deadlock in the smb redirector and also adds new 
   functionality that permits the AFS SMB server to be given a longer 
   timeout than is normally the case.  New functionality has been added 
   to configure these additional LanmanWorkstation\Parameter values.
   (This functionality has been backported to XP SP3 and is scheduled
   to be released on June 5th.)
   
 * The BackConnectionHostNames registry value configuration was broken 
   when dynamic re-establishment of Netbios Name registrations was added.
   Restore it.

 * Hidden vos commands are revealed.

 * If the "DisableLoopbackCheck" registry value is set, do not unset 
   it during the same service session.

 * Reorganize code that prevents multiple Store operations to the same 
   File.

 * Modify IsPathInAfs test in Explorer Shell Extension and fs.exe to 
   permit broken symlinks to be treated as being in AFS.

 * Fix vos commands that output 64-bit integer values.

 * Cygwin Import Libraries are provided in the SDK for all OpenAFS DLLs.
   Permits building cygwin applications against OpenAFS libraries. 

 * OpenAFS Release Notes, Administrator Guide and User Guide now installed 
   as Windows HTML Help (.CHM) files.  

 * NSIS installer does a much better job of cleaning up files left 
   over from previous installs.

 * Fix RT#124787, a race condition between "fs flush <dir>", "fs flushvolume", 
   or "fs flushall" and on-going directory operations that can result in 
   afsd_service.exe crashing.

 * Add support for DNS SRV records in place of AFSDB records.  
   _afs3-vlserver._udp.<cellname>.  Priority field is used.  
   Weight is currently ignored.

 * Add a method of specifying Client CellServDB information within the 
   registry that can be used to either override the CellServDB file or 
   force the use of DNS lookups for a given cell.

   The registry schema is as follows:

     HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\
       "LinkedCell" REG_SZ "[cellname]"
       "Description" REG_SZ "[comment]"
       "ForceDNS" DWORD {0,1}

     HKLM\SOFTWARE\OpenAFS\{Client,Server}\CellServDB\[cellname]\[servername]\
       "HostName" REG_SZ "[hostname]"  Default: [servername]
       "IPv4Address" REG_SZ "[address]"  Used only if gethostbyname() fails.
       "IPv6Address" REG_SZ "[address]" <future>
       "Comment" REG_SZ "[comment]"
       "Rank" DWORD "0..65535"  Default: 0
       "Clone" DWORD "{0,1}" <future: server only>
       "vlserver" DWORD "7003" <future>
       "ptserver" DWORD ... <future>

   ForceDNS is implied non-zero if there are no [servername]
   keys under the [cellname] key. Otherwise, ForceDNS is zero.
   If [servername] keys are specified and none of them evaluate
   to a valid server configuration, the return code is success.
   This prevents failover to the CellServDB file or DNS.
   
   Only one of "HostName" or "IPv4Address" is required.  "HostName"
   is optional if [servername] is the host name.

 * Extend registry based CellServDB functionality to 
   afsconf_GetCellInfo() interface used by aklog and many 
   other command line utilities.

 * libafsconf.dll moved from OpenAFS\Client\Program to OpenAFS\Common
   as it is now used by both client and server components.

Since 1.5.58
 * PriorityClass of afsd_service.exe process raised to 
   "High" to match the priority of the system services 
   that are waiting for afsd_service.exe to respond.
   Configurable using [TransarcAFSDaemon\Parameters]
   DWORD "PriorityClass" value.  (use values from Win32
   API)

 * Change error reporting of CM_ERROR_BADSHARENAME to
   use the correct error (NTSTATUS_BAD_NETWORK_PATH).
   The error we were sending would cause the smb 
   redirector to drop the connection.

 * Change the error reporting of CM_ERROR_ALLOFFLINE
   and CM_ERROR_ALLDOWN to return NTSTATUS_PATH_NOT_FOUND.
   Previous we would return Bad Network Path which 
   would cause the smb redirector to cancel all file
   handles on the same "share".

 * A debug version of afsd_service.exe can now be used
   with a release version of afsrpc.dll (or vice versa).
   afsrpc.dll now implements and exports xdr_free().

 * Removed all unprotected OutputDebugString() calls as
   those calls can block for up to 10 seconds if there 
   is contention for the globally shared memory or if
   the debugger fails to read data in a timely manner.

 * Visual Studio 2008, WDK 6.1 and DDK 6.1 are now 
   supported for builds.  The production releases still
   use VS 2005.

Since 1.5.57
 * Discovered that XP/2003 supports a variant on recvmsg 
   called WSARecvMsg.  The input parameters are mostly 
   compatible with the Posix recvmsg.  This will save 
   multiple memcpy() calls per received packet.

 * Discovered that Vista/2008 supports a variant on 
   sendmsg called WSASendMsg.  The input parameters 
   are mostly compatible with the Posix sendmsg.  This 
   will save multiple memcpy() calls per sent packet.

 * Exporting rx_ReadProc32, rx_WriteProc32 from afsrpc.dll 
   and utilizing rx_ReadProc32 from within the cache manager 
   as part of the FetchData operations.  rx_ReadProc32 and 
   rx_WriteProc32 are optimized versions of rx_ReadProc and 
   rx_WriteProc for use when reading 32-bit integer values.

 * The Network Identity Manager AFS provider now reports 
   progress events.  These provide a better user experience 
   with Network Identity Manager 2.0.

 * [RT 124293] A race condition exists which permits the 
   scp field of the an smb_fid_t object to become invalid 
   after a request on the smb_fid_t has begun.  Remove the 
   race by protecting all access to the scp field with the 
   smb_fid_t mx mutex and obtaining a local reference on 
   the cm_scache_t object for the length of the request.

 * [RT 124276] If the vldb is out of sync with the contents 
   of the file servers, afsd_service will retry too many 
   times when a file server reports a volume as not being 
   present.  Now if the list reported by the vldb is the 
   same as the previously seen list, then the retry is 
   aborted.

 * [RT 124276] Read-only volume failover was broken in 
   1.5.53 whenever accessing a volume results in VNOVOL or 
   VMOVED.

 * [RT 124276] Prior to 1.3.70 the volume server reference 
   list was not reference counted and would be prematurely 
   freed while in use.  When reference counting was added 
   in 1.3.70 a bug was introduced that could result in 
   service reference list corruption.

 * Add Windows Application Event Log warning messages for 
   "Client SMB MPX value too large" and 
   "Client SMB Buffer Size too small".

 * Renaming of files across directory boundaries would result 
   in an invalid handle error when attempting to access the 
   files after the move.

 * Fix the handling of Tran2 Set Path Info RPCs.  Do not fail 
   when a smb file descriptor cannot be found.  The whole point 
   of using a Path Info function is because an smb file 
   descriptor wasn't allocated.

 * More edge cases in which dynamic addition of Freelance 
   root.afs entries would get the wrong FID or where the 
   root.afs directory would not be refreshed.

 * Buffer overflow could occur if the workstation cell name was 
   longer than 64 characters.  Crashes could occur in 
   afscreds.exe, afslogon.dll, and afsd_service.exe.

 * VNOSERVICE and VOFFLINE errors were leaking and were exposed 
   to the smb client.

 * Improve consistency of the bad data version value used for 
   cm_scache fields.

 * Log file server uuid values as part of the cm_server object 
   when available.  Dump the cm_server object list in response 
   to "fs memdump".

 * Optimize the performance of resetting access control lists 
   when tokens are set or removed.

 * Fix vos.exe so that it can work with very large volume 
   identifiers

 * Remove symlink recursion tests and increase max symlink count 
   to 64 from 16.

Since 1.5.56
 * Undo the "UAC manifest fix" applied to afs_config.exe.

 * Fixed some additional race conditions in the Rx library

 * Ensure that Freelance allocation of vnodes follow 
   the AFS convention of odd vnodes are directories 
   and everything else is an even vnode.

 * Add Freelance logic to mount point and symlink 
   evaluation functions.

 * Enhance smb_ParseASCIIBlock() so that it can handle
   all of the STRING formats defined by the CIFS Technical
   Report 1.0.  

 * Validate the output of smb_ParseASCIIBlock() in all
   callers.  Return CM_ERROR_BADSMB if the STRING field
   cannot be parsed.  CM_ERROR_BADSMB will cause the
   contents of the packet to be logged.

 * If multiple SMB Raw Write operations were taking 
   place at the same time, there could be data corruption
   because unique event objects were not generated for 
   each Netbios receive operation.

Since 1.5.55

 * Fix a race in rxi_WriteProc which could result
   in a freed or in use packet being passed to 
   rxi_PrepareSendPacket if the call is currently
   in a transmit wait state.  Passing a freed packet
   will result in a panic.  Passing an in use packet
   might result in a call failure or perhaps data
   corruption.

 * Kerberos referrals broke the NIM AFS provider 
   logic for automated configuration of the workstation
   cell with a new identity.  Now if a referrals 
   response is returned, assume the cell belongs to the
   new identify.

 * The interface to buf_CleanAsync() and 
   buf_CleanAsyncLocked() has been changed to permit 
   errors to be propagated to the caller.  

 * During an enumeration of a directory, if the btree
   has not yet been generated, return an error.

 * When mapping CM_ERROR values to NTSTATUS values
   do not treat 0 as an error.

 * Avoid a cm_scache_t object reference count underflow
   when renaming via the SMB interface.

 * smb_IsMask() can be called with a non-8.3 input buffer.
   Separate smb_IsMask() into smb_Is8Dot3Mask() for 8.3
   input buffers and smb_IsMask() for non-8.3 input buffers.

 * Add cm_buf_t error state to the "fs memdump" output.

 * pioctl drive mapping checks need to take into account
   Global Drive Mappings and SUBST assignments.

 * Rename operations should not be permitted if the 
   destination name collides with an existing name that
   differs only by case UNLESS the collision is the object
   that is being renamed.

 * Increase the maximum number of background daemon threads
   to 64.  Leave the default at 4.

 * Set the FILE_UNICODE_ON_DISK flag when reporting the 
   file system characteristics.

 * Report the maximum component name length as 255 even
   for Unicode as part of the file system characteristics.

 * Add RX debugging information to "fs memdump" output
   in checked builds.

 * Refcount the cm_server_t objects returned by
   cm_FindServerByIP()

 * Fix the UAC manifest applied to afs_config.exe

 * rxdebug now indicates if any RX_CONN flag is set
   not just some of them

 * Change the mapping of CM_ERROR_WOULDBLOCK from 
   NTSTATUS_LOCK_NOT_GRANTED to NTSTATUS_CANT_WAIT

Since 1.5.54

 * Includes #GCO Public CellServDB 07 Nov 2008

 * Prevent separate cell entries from being created for
   "foobar.org" and "foobar.org." which can happen if
   an AFSDB record is found instead of a CellServDB 
   record.

 * Fix a panic when checking the SMB virtual circuit list.  
   If a VC is destroyed while the list is being checked 
   it is possible for the next object to become invalid.

 * Increased logging to afsd_init.log of netbios name 
   related errors.

 * Fix a DNLC panic when a UTF-16 string that cannot be 
   converted to a utf-8 string is provided as a filename.

 * In the NSIS installer, delete old afslogon.dll/afscpcc.exe 
   from %windir%\system32 and fix the WinLogon\Notify\AfsLogon 
   location to refer directly to the new location in 
   %programfiles%\openafs\client\program

 * Be more aggressive about responding to Netbios 
   NRC_BRIDGE/NRC_NOWILD errors.  Rebind the Netbios name to 
   the LANA within the listener thread without performing a 
   delete and reset.

 * Use the WNet Enumeration of all connected drives as a filter 
   in the pioctl GetIoctlHandle operation.  This should remove 
   the last situations in which there was a delay when opening 
   an Explorer Shell properties menu with the AFS Shell 
   Extension.

 * Fix the parsing of CellServDB in Windows to handle linked 
   cells as is done on Unix.

 * Add support for linked cells to the cache manager,
   netidmgr afs provider, and aklog.  When two cells are
   linked, a volume lookup that fails in one cell is looked
   up in the other.  This can be used for construction of 
   test cells that override specific volumes; cell renaming;
   and cell splitting.

 * Increase the length of cell and realm names that can 
   be entered in the NIM AFS provider.

 * Actively prevent a 32-bit afsd_service.exe from 
   executing in the WOW64 environment.

 * Protect against invalid IP address values in CellServDB
   parsing.

Since 1.5.53

 * Fix an Rx panic caused by an attempt to double free
   an rx_packet when an error occurs while writing the
   packet.

 * Fix an exception cause by a race condition when 
   processing multiple file change notifications registered
   by the Explorer Shell or other applications.

 * Modify the Rx debug interface to use var_args and 
   permit Rx debugging to be used on 64-bit Windows 
   which previously had it disabled.

Since 1.5.52

 * Automated lock hierarchy enforcement added to the
   osi base lock package.  This permits detection of lock
   violations that might but often does not result in 
   deadlocks.  Enabled by default in the checked builds.
   May be activated in release builds via the registry
   value "LockOrderValidation" in the TransarcAFSDaemon
   Parameters key.

   When a violation is detection an assertion is thrown
   generating a dump file and a message indicating which
   locks were obtained out of order is logged to the 
   Windows Application Event Log.

 * Attempt to ensure that a dump file will be generated
   whenever the service panics.

 * Removed unused lock cm_bufGetMutex

 * Many out of order lock acquisitions were detected
   using the lock order verifier and were corrected.

 * A race condition within the function path cm_GetSCache() ->
   cm_GetNewSCache() -> cm_RecycleSCache() permitted a 
   cm_scache_t object to be issued simultaneously to two 
   threads.   This would eventually result in a panic due 
   to the resulting undercount.

 * Fix interpretation of the empty string as the ioctl path
   to mean the current directory.  "fs lsm", "symlink list", 
   etc. now return a "not a ..." error instead of "does not
   exist".

 * Add SMB_STRF_SRCNULTERM flag to smb_ParseStringXXX functions.
   to indicate that the input string is a nul terminated string.
   Assign ti when input strings are nul terminated instead of 
   length counted. (from 1.5.50)

 * The CIFS protocol specification for handling NT_TRANSACT_CREATE
   does not match the observed behavor.  The 'nameLength' is specified
   in bytes, not in characters.  Fix the implementation to match
   the observed behavior. (1.5.50)

 * The cm_HaveAccessRights() attempt at deadlock avoidance by 
   calling lock_TryRead() on the parent directory cm_scache_t rw-lock
   does not avoid the deadlock.  Avoid the deadlock by enforcing the
   lock order of the lowest vnode first.  Then remove the infinite 
   loop avoidance in cm_SyncOp() that was returning an unwarranted
   access denied error.

 * File Server lock synchronization was not propertly enforced in 
   smb_ReceiveNTCreateX() and smb_ReceiveNTTranCreate().

 * Another change to the short name generator.  Plus and equal are 
   not valid characters in a short name according to the 
   RtlIsNameLegalDOS8Dot3() in kernel mode.  They have been replaced
   with left and right brace in order to minimize the impact.

 * Fix a race condition that can result in a heap overwrite error if
   the number of known file servers or vldb servers increases during
   the preparation phase of a cm_CheckServers() operation.

 * If the SMB Authentication would fail, a LSA allocated memory 
   buffer would be leaked.

 * Rewrite the cm_cell_t allocation code to permit cell vldb server
   lookups to fail without leaking cm_cell_t objects.

 * Add "fs setcrypt auth" to permit the use of rxkad authentication
   only mode.  Not documented.

 * Initialize Rx mutex and lock objects once and only once as part of
   the DLL load process.

 * When reporting the Volume Name "AFS" do not include the terminating
   NUL in the reported length.  
 
 * Refresh Volume Location information upon receipt of a VNOVOL error
   from the file server and avoid a race condition that prevented the
   updated volume location information from being destroyed immediately
   after it was acquired.

 * In response to VNOVOL and VMOVED, remove the volume id from the server's
   volume list so that the volume's up and down status is not affected 
   when the machine is shutdown.

 * Many fixes to the Rx stack related to packet allocation and deallocation;
   Rx debugging; Rx packet tracking across queues; and Rx packet loss caused
   by overwriting the rx_call currentPacket pointer when it was already 
   referring to a packet.  

 * Rx Jumbograms are now off by default.  To re-enable, add "RxJumbo" = "1"
   in the TransarcAFSDeamon/Parameters key.

 * "fs flush", "fs flushvolume", and "fs flushall" now destroy locally 
   constructed directory B+ trees for flushed directory objects.

 * When cached pages for directory objects have been locally modified, do
   not permit the locally modified pages to be mixed with pages subsequently
   obtained from the file server.  Doing so can result in directory 
   corruption.

 * If Ubix errors are received in response to VL_xxx operations record
   the error on the server instance and permit the operation request to 
   be retried in case another server is able to answer the query.  We have
   seen at least one instance of a cell in which a VL server returned 
   UNOQUORUM (possible database truncation) when the other servers were
   responding successfully.

Since 1.5.51
 * Optimize the assignment of write errors to dirty 
   buffers by applying the first error to all buffers 
   belonging to the same object instead of trying each 
   buffer in turn.

 * Add resource version info to xstat_fs_test and xstat_cm_test
 
 * Fix a delete vs delete[] usage problem in the WiX custom 
   handler dll
 
 * When creating a mount point, validate the volume name to 
   make sure it exists.

 * When processing a mount point, pay attention to VL_BADNAME 
   errors.

 * Remove a race condition from cm_volume_t recycling that could 
   have resulted in two threads making use of the same object at 
   the same time.

 * Fix a reference count leak on smb_vc_t objects introduced in 
   1.5.50.  Permit the objects to be freed once again.

 * Fix the output of "fs memdump" for smb_vc_t objects that was 
   broken in 1.5.50.

 * Add routines for debugging smb refcount errors.

 * When shutting down the cache manager, do not permit AFS 
   callback requests to be processed. Doing so can result in 
   invalid memory accesses.

 * Do not recycle a deleted cm_scache_t object unless the 
   refCount is 0.  Doing so can result in a deadlock and 
   invalid memory access.

 * Minimize the number of RPCs sent to the file server for 
   directory bulk status requests. Only request status for 
   objects that we are actually reporting on and only if we 
   do not already have status for the object and then make 
   sure that we return as much as we can at once via the SMB 
   interface but not so much that it exceeds the RDR timeout

 * Enable bos restricted operations

 * Fix the creation of submounts used by the afscreds and afs_config 
   drive mapping tabs.
 
 * Avoid repeated retries when attempting to obtain access rights.  
   If the access rights were obtained successfully but do not 
   satisfy the request, fail the request.

 * Fix error code reported when attempting delete a file on a 
   readonly volume or one that is marked with the readonly DOS 
   attribute.

 * Fix a short name truncation bug.

 * Fix a heap corruption error when reading the CellServDB file 
   location.

 * Fix directory listings when the directory has "lookup" permission 
   only and the directory has never been previously accessed.

 * Fix several tracking issues for cm_scache_t rw locks.

 * Ensure that all buffers used to store cell names are the
   correct length.

 * Restore Windows 2000 support (broken in 1.5.50).   
   LOCALE_INVARIANT was introduced in XP.

 * Add RxUdpBufSize registry value.  Default is 256K.

 * Avoid multiple deadlocks between cm_freelanceLock and cm_scacheLock

 * Do not include trailing NULs when returning file names in 
   a directory search response.

 * Prevent RX from crashing the afsd_service by attempting
   to allocate to many buffers at once when memory is fragmented.

 * Avoid a deadlock when freeing cm_server_t objects

 * Optimize search cell file.  Do not search to end of file
   if the cell has already been found.

 * Remove explorer shell extension delay on local disks.

 * Automatically pre-allocate 64 Rx packets per thread

 * Add smb lock requests and stat cache lock allocations to
   the output from fs memdump.

Since 1.5.50
 * Fix Large File Support (broken in 1.5.50)

 * Fix "fs setacl" (broken in 1.5.50)

 * Fix pioctl path evaluation of drive letter
   mapped paths. (broken in 1.5.50)

 * Fix installation of VC8 Runtime library in the EXE
   Retail installer. (broken in 1.5.50)

 * Avoid a potential deadlock during Volume Location 
   updates if all the file servers are unreachable at 
   the time the VL update succeeds.

 * Ensure that rx connection object is references 
   during VL RPCs.

 * Fix propagation of over quota errors during cm_FSync()
   calls.

Since 1.5.39 [1.5.50 released 16 July 2008]
 * There were no releases between 1.5.39 and 1.5.50.

 * FIRST RELEASE WITH UNICODE SUPPORT

 * Pioctl interfaces to the cache manager refactored
   to provide layering between the SMB specific code
   and the general purpose ioctl operation.

 * Garbage collect dead SMB virtual circuits as soon
   as they are no longer being referenced.  This avoids
   problems with outstanding locks not being dropped
   when the virtual circuit becomes invalid.

 * Remove the IBM Administration Reference documentation
   and replace it with the OpenAFS Command Reference 
   Manual.  

 * Avoid calling rx_SetDeadTime and rx_SetHardDeadTime
   functions each time a connection is about to be used.
   Do not hold a lock on the rx connection object while
   it is being selected.  This avoids a race between 
   threads attempting to set the timeout values and 
   removes a bottleneck that was hampering performance.

 * Ensure that the smb directory attribute is set 
   for all directory objects.

 * Add vs2008 support to the NSIS installer scripts

 * Replace the VC Runtime EXE installer with the MSI
   installer in the NSIS installer scripts

 * Properly delete the VC7.1 C Runtime libraries.   

Since 1.5.36 [1.5.39 released 23 June 2008]
 * There were no 1.5.37 or 1.5.38 releases for Windows

 * Use Visual Studio version of vsprintf() instead of 
   home grown version that resulted in linking conflicts

 * Prevent buserver, ptserver and fileserver from crashing
   due to non-portable strftime() format strings.

 * Prevent server crashes caused by rx_InitMorePackets()
   being called prior to allocation of all of the required
   synchronization objects.

 * Fixed a memory leak in the internal B+ tree directory.
   For objects with non-8.3 names, the 'longname' memory
   allocation would be leaked during object deletion or
   tree destruction.

 * Fixed a memory leak in the background daemon cm_CheckServer() 
   routine.  The memory allocated to store the server list
   was not freed.  

 * Obtain a missing lock around a call to cm_RemoveSCacheFromHashTable(). 

 * Correct an abstraction layer violation. cm_scache_t objects 
   should be marked deleted in cm_Unlink() and cm_RemoveDir() and not 
   in smb_CloseFID(). Cleanup of deleted cm_scache_t objects should be 
   performed in cm_ReleaseSCache() when the reference count hits zero. 

 * Prototype cm_AdjustScacheLRU() and re-implement it using osi_QAddH().

 * Do not reference 'smb_logp' in cm_vnodeops.c.  Use 'afsd_logp'
   instead.

 * Ensure that scache object 'nextp' pointers are set to NULL
   when the objects are removed from the hash table linked list.

 * If a dirty buffer cannot be written to the file server because
   of an access denied error, mark the error in the cm_buf_t
   structure so that it can be dealt with instead of retrying 
   forever.

Since 1.5.35
 * Update CellServDB file to GCO Public 23 Apr 2008

 * Fix a cm_buf_t reference count leak when attempts to write 
   dirty buffers to the file server from within cm_IncrSyncer() 
   fail.

 * Prevent udebug from crashing.

 * Activate RX Idle Timeouts.  If the file server is busy for more 
   then 30 seconds, attempt to failover to another server without 
   marking the busy server down.

 * Another VNOVNODE issue fixed.  When writing a dirty buffer
   to the file server, if VNOVNODE is received, mark all buffers
   as invalid without further attempts to contact the file server.
   
 * Improved performance on high latency links.

Since 1.5.34
 * Fix a bug preventing the re-initialization of the Freelance 
   root.cell contents when a change is made.

 * Properly return CM_ERROR_NOSUCHFILE when removing a symlink 
   or mountpoint that does not exist.

 * aklog, afscreds, integrated logon, and the NIM AFS credential 
   provider will all attempt to first acquire an afs service ticket 
   using the service principal: afs/<cell>@<CLIENT_REALM>.  This 
   permits the proper detection of the cell realm when multiple 
   realms are local realms for the cell.

 * Fix fs commands that re-use file identifiers instead of forcing 
   a path evaluation each time.  This was broken in 1.5.33 when 
   hashes were added to the cm_fid_t structure.

 * Fix symlink and mountpoint removal and then recreation when 
   applied to the fake Freelance root.afs volume.

 * Fix SMB change notifications for the Freelance root.afs volume 
   when creating or removing symlinks and mount points.

 * Force DNS AFSDB cell name lookups to lower case if the DNS 
   response is not all lower case.

 * 32-bit installers are now built with Visual Studio 2005

 * All vos.exe commands now support the "-noresolve" option which 
   forces the display of IP addresses instead of DNS hostnames.

 * No longer hold the cm_volume_t mutex across VL_xxx RPCs.  This 
   permits operations that do not require up to date volume 
   location data to continue while waiting for the RPC to complete.  
   Multiple threads waiting for a volume location update on the 
   same volume set no longer trigger multiple calls VL_xxx RPCs.

 * Volume server references could be marked offline and never 
   reset to the not_busy state.  

 * Further optimizations to cm_Analyze() retry logic for 
   CM_ERROR_OFFLINE and CM_ERROR_ALLBUSY error handling.

Since 1.5.33
 * Optimize the DNLC by applying Interlocked incrementing for 
   statistics gathering and enabling greater use of read 
   locks instead of write locks

 * Further optimize osi_Log operations when logging is disabled 
   by removing the need for the function call overhead.

 * Add a new lock operation, Convert Read to Write, and apply 
   it throughout the cache manager.  This function provides a 
   fast transition from a read lock to a write lock when the 
   caller is the only reader.

 * Fix a timeout problem when opening files or request locks 
   when the file in question is already write-locked by another 
   machine.

 * Fix a deadlock that can occur if "fs flushall" is executed 
   while applications are writing to AFS.

 * The cache manager would re-read data from the file server 
   after file truncation when obtaining buffers to write to the
   same file until the first StoreData RPC completes.

 * Fixes error handling in cm_NTCheckDelete()

 * Replaces cm_scache_t mutex with a read-write lock permitting 
   additional parallel access.

 * Prevent an cm_scache_t reference undercount when evaluating 
   symlinks containing @sys when none of the active sysnames 
   match any of the available targets.

 * Prevent the leak of a cm_scache_t rwlock if associated symlink 
   is too long.

 * Set RxMaxMTU registry value to 0 which is the equivalent of 
   it not being set at all.

 * Reduce contention between BkgDaemon threads

 * Prevent CM_SCACHEFLAG_ASYNCSTORE from being reset on a write 
   failure

 * Convert reference counts on cm_volume_t, cm_conn_t, cm_server_t 
   objects to use the faster Interlocked operations which avoid 
   the need to hold write locks.

 * Permit rxkad errors other than RXKADEXPIRED to be treated as a 
   non-fatal error that forces a retry against another service 
   (if available.)

 * Fix a crash in smb_WriteData that would occur if an application 
   opened the file as read-only and then sent a write request anyway.  
   (RT 88731)

 * Respond to VL_NOENT errors by removing cm_volume_t object and
   setting it to be the first object to be recycled.  This is not 
   only a more efficient use of resources but it also prevents 
   repeated attempts to query the VLDB server for the volume from 
   the CheckOfflineVolumes routine.

 * Change the defaults for RxEnablePeerStats and RxEnableProcessStats 
   back to "on".

 * A first cut at a cache manager statistics monitor that can be used 
   to determine the necessary cache parameters to support the 
   application working set.

   Off by default, the performance package can be activated by setting 
   "daemonPerformanceTuningInterval" in the service Parameters key.  
   As with the other daemon interval values the unit is in seconds.

   At service start and each succeeding interval the cache manager 
   will write statistics to %TEMP%\afsd_performance.txt showing the 
   relative usage of cm_scache_t, cm_volume_t and cm_buf_t objects.  
   The FID statistics keep track of all FIDs seen by the cache manager 
   during the service session whether or not they are backed by any 
   live object in the cache.

   These statistics are not stored in the cache file.

 * Fix windows event logging broken in 1.5.33.

 * Eliminate multiple calls to syscfg_GetIFInfo() which is used to 
   populate the IP address list for WhoAreYou/TellMeAboutYourself RPCs
   as well as when computing the server preference list.

 * When merging status after a StoreData operation prevent data 
   buffers from being discarded that we would prefer to keep.  
   (Bug introduced in 1.5.33.)

 * Modify installers the set RxMaxMTU to 0.

 * In the AFS Configuration Panel on the Advanced->Miscellaneous 
   dialog, the Background Daemons and Service Threads fields were 
   swapped.

 * Increase default number of Background Daemons to 4.

 * Remove cm_volume_t reference from the cm_scache_t object.  Permit 
   cm_volume_t objects to be recycled.

 * Windows Error Reporting has shown a number of previously unseen 
   stack corruption errors in the Explorer Shell extension.  The 
   minidump appeared to indicate the problem was in ParseAcl().  
   Reviewing the function it appears that numerous things could go 
   wrong if invalid input was provided.  Added error checking and 
   data validation.  Hopefully this will address the problem.

 * Windows Error Reporting has shown crashes in aklog as a result
   of krb5_cc_get_default() failures not being handled.  They are
   now handled.

Since 1.5.32
 * The Rx library used a 32-bit type for sockets which was
   truncating the socket value on 64-bit Windows.  This 
   was not an issue in general but became an issue when 
   Rx "hot thread" support was enabled.

 * Enable "Rx hot thread" support which ensures that while
   a received message is being processed that a new thread
   is allocated to receive the next message to arrive.

 * Increased parallel access to smb_rctLock protected 
   resources.  Reduce the number of times that the 
   lock is dropped and re-obtained.

 * Increased parallel access to cm_scache_t and cm_buf_t
   resources by replacing the previous global write lock
   protected reference counting mechanism with one built
   upon the use of InterlockedIncrement and InterlockedDecrement.

 * Implement a hash for cm_fid_t to reduce comparison costs
   and inline the cm_FidCmp() function.

 * Redefine the BUF_HASH and BUF_FILEHASH in terms of the cm_fid_t 
   hash which has a better distribution

 * Modify cm_ConsiderPrefetch to evaluate the amount of data in 
   the most recent read request instead of the next chunkSize.  
   cm_chunkSize can be dozens or hundreds of buffers.  As a result
   too much time is spent performing the evaluation.

 * Fix the usage of cm_scache_t bufCreateLock.  The purpose of this 
   lock is to prevent the creation of new buffers while a truncation 
   is being performed.  All references to bufCreateLock have been 
   removed except in two places:
   i.  a write-lock surrounding the function that calls buf_Truncate()
   ii. a read-lock within buf_GetNewLocked() that actually allocates 
       new buffers

 * Modify the CM_CONFIG_DATA_MAGIC value to include a 
   CM_CONFIG_DATA_VERSION number which value be used to force the 
   replacement of the cache file contents when incompatible changes 
   are made between releases.

 * CM_SCACHESYNC_ASYNCSTORE should not be ordered by 
   cm_SyncOpCheckContinue

 * Avoid calls to multi_Rx if nconns == 0

 * Modify smb_WriteData to perform background writes based upon the 
   crossing of 'smb_AsyncStoreSize' boundaries instead of cm_chunkSize 
   boundaries.  This will slow down writes from the SMB interface but 
   will avoid the risk of the CIFS client disconnecting from the AFS 
   client SMB server.  The default value for 'smb_AsyncStoreSize'
   is 128K.  

 * SMB AsyncStore functionality is on by default but can be 
   disabled using the registry value:

     HKLM\SOFTWARE\OpenAFS\Client DWORD "EnableSMBAsyncStore"

   smb_AsyncStoreSize can be adjusted with 

     HKLM\SOFTWARE\OpenAFS\Client DWORD "SMBAsyncStoreSize"

   The value must be a multiple of the buffer block size 
   and cannot be larger than the chunk size.

 * Do not leak a read lock on the bufCreateLock

 * Do not forget to call cm_SyncOpDone if cm_SyncOp succeeded
   in cm_CheckNTOpen()

 * In MergeStatus, only remove clean buffers from the hash table.

 * Avoid the need to update the data version number on each buffer 
   associated with a scache when MergeStatus is called after a 
   StoreData by maintaining a range of valid data versions as part 
   of the cm_scache_t object.

 * Profiling shows large numbers of blocked calls in cm_HoldSCache
   from cm_HaveAccess...   We can safely avoid the call so do so.

 * Add "ForceAnonVLDB" service parameters registry value.  Default
   is 0.  Set to 1 to force the use of anonymous connections to
   the VLDB servers.  This is useful when using krb5 tokens and the
   VLDB servers have not been updated.

Since 1.5.31
 * Do not attempt to search for a cell name that matches one
   of the special share names "svrsvc", "wkssvc", "ipc$"

 * Issue permission denied warnings from afscreds if attempts
   to modify the startup shortcut or registry values fail.
   This happens under Vista UAC.

 * Do not return BAD_NETWORK_ERROR in response to attempts
   to create directories or files in the Freelance root.afs 
   volume.  Instead return ACCESS_DENIED.

   BAD_NETWORK_ERROR was being sent because the Freelance cell 
   has no vldb servers.  Short circuit the error by testing for 
   freelance in CreateFile and MakeDir

 * Treat srvsvc, wkssvc and ipc$ as pseudo files.  Permit them 
   to be opened and closed without actually existing.  For now 
   we treat them as equivalent to ioctls but this should 
   eventually change.

 * Fix RAP processing to not reject names that exist in the 
   root.afs volume for GetInfo queries.

 * In afscreds, use GetUserNameEx to obtain the username instead 
   of searching the registry.  Try UserPrincipalName first and 
   fallback to SamCompatible.

 * Stop linking to rpcns4.dll which is no longer supported on 
   Vista and 2008.  TaAfsAdmSvr is going to have to be 
   redesigned.

 * When processing dfs referral requests, if the sharename is 
   a partial match and therefore will not be added automatically 
   to the share list, make sure that we return an error.

 * Avoid a crash when logging VNOVOL errors returned from 
   cm_CheckOfflineVolume.

 * When trace logging is disabled, do not permit osi_LogSaveString 
   to perform the string copying.  By disabling the string 
   copies, throughput of the service more than doubles.

 * when checking offline volumes, update the cell vlserver info
   if required and don't perform a check if there are no servers
   in the list.

 * Add volumes in the vl_unknown state to the list of objects to be
   checked as offline or retried when an allbusy error is received
   in cm_Analyze.

 * Add support for configuring rx_ExtraPackets

 * All calls to buf_Get() must be protected by a read lock on the
   cm_scache_t bufCreateLock

 * When a MergeStatus results in the invalidation of buffers, do
   not remove buffers from the hash tables that have active references.

 * Upon receiving a VNOVNODE error from a file server, be sure to
   invalidate any file locks.  This prevents on-going attempts to
   extend the locks and will cause subsequent attempts to access
   the file to receive a invalid handle error.

 * Prevent smb_ReceiveTran2SearchDir from adding terminating NULs
   to directory entry names.

 * No longer discard tokens when RXKADBADTICKET error is received.
   Instead return STATUS_NO_KERB_KEY error.

Since 1.5.30
 * Add "cmdebug <host> -cellservdb" option and client service 
   implementation permits querying the vldb server for each 
   cell known to the cache manager.  The output is in a form
   that permits it to be used as a CellServDB file.

 * Add Symlinks->Show option to the AFS Explorer Shell Context
   Menu

 * Prevent buffer overrun in the cmd parser used by command
   line tools such as fs, vos, pts, etc.

 * The path ioctl operations have several issues:

   (1) the specified path for "fs examine, whereis, whichcell, 
       flush" and so always has the follow symlinks and mount 
       points semantics.  This makes it impossible to determine 
       what the FID of a symlink or mount point is.

   (2) "fs examine" out is not the result of a single pioctl 
        operation but is actually the combined output of half 
        a dozen operations.  Path evaluation is an expensive 
        operation.  It would be faster if the caller could
        evaluate the FID first and then perform all of the 
        rest of the operations by specifying the FID instead 
        of the path.

   (3) fs output reports all objects as files.  By adding a 
       GetFileType pioctl more informative output can be provided 
       that indicates what type of object the path evaluates to.

   (4) the Windows fs command includes a number of commands that 
       do nothing but exist only because the Unix cache manager 
       supports them.

   A new extendible data structure cm_ioctl_query_opts_t has been 
   added which can be optionally specified with pioctls that do not
   already require input data.  The first two fields of this structure 
   are 'literal' and 'fid'.  The literal field is used to indicate
   whether the last component of the path should be evaluated following 
   symlinks and mount points.

   The fid field permits a fid to be specified.

 * A new GetFileType pioctl has been added. The type of objects are 
   now output.

 * A new "-literal" option is available for "fs examine, flush, whereis, 
   and whichcell.

 * Unimplemented fs commands have been removed.
   
 * There are circumstances where a volume object is being accessed and 
   the volume is marked indicating that the volume location information 
   is out of date but where it is also pointless and perhaps dangerous 
   to block waiting for the rpc to complete.  One example is when 
   processing the cmdebug requests.  If we know that we are not going 
   to use the volume object to contact a server, then we can now set 
   the CM_GETVOL_FLAG_NO_RESET flag.

 * when moving up in the directory tree we search the recorded fid list 
   to find a matching fid that we have already crossed.  we must also 
   reset the fid count based upon what we discover.

 * Cell names are published as share names.  As a result they are searched
   for as part of the Dfs Referral evaluation.  If share "foo" can not be
   reached, the CIFS client will fallback to searching for "fo" as well.

   Since the freelance client automatically adds a symlink for prefixes
   of the cell name, this results in both "foo" and "fo" being added to
   the freelance root.afs volume.  This patch prevents that negative side
   effect.

 * when tracking Kerberos credential cache names be sure to include 
   the cache type prefix.

 * Microsoft has assigned OpenAFS a network type value

 * The algorithms that were used to produce a Netbios name were broken.
   The name that was produced when the hostname was too long was in fact
   longer than the maximum permitted netbios name.

   Also, the "NetbiosName" value was not used as a suffix when the loopback
   adapter was not installed.  The hardcoded string "AFS" was used instead.

 * avoid another deadlock during server probes initiated by the ipaddr 
   change daemon thread.

 * prevent the afs client service from crashing when fs newcell is 
   executed

 * In the NSIS installer, move afslogon.dll, afscpcc.exe and afs_cpa.cpl 
   to \Program Files\OpenAFS\Client\Program in order to get them out
   of the %windir% tree

 * EDQUOT == WSAEDQUOT.   Define it in the right places so that in the 
   end STATUS_OUT_OF_QUOTA can be returned to the application.

 * Modify the search order for determining the location of CellServDB
   and other client configuration files.

   1. AFSCONF environment variable
   2. registry setting
   3. NEW - use All Users\AppData\OpenAFS\Client only if CellServDB exists
   4. use Program Files\OpenAFS\Client

 * More Vista Power Management tweaks

 * Close a socket handle leak in the kauth routines

 * Don't hold a cm_volume_t mutex across RPCs

 * Convert downTimes to use clock time instead of relative times.
   This permits correct comparisons with expiration times.

 * Update the Wix installer to properly install the prerequisites 
   for translate_et.exe

 * Do not translate pioctl paths as they are not converted by the
   file system stack.

 * Implement multi_rx probes for checking whether servers are up
   or down.  Turns probes into a constant time activity instead 
   of one dependent upon the number of known servers.

 * Treat VNOVOL errors the same as VOFFLINE.  Do not force an
   update of the volume location data and retry.  VNOVOL means
   that the volume could not be attached.  It doesn't mean that
   the volume is not on this server.

 * Modify the Explorer Shell AFS Context menu Symlink->Add to
   not attempt to validate the target path.   There is no reliable
   method of doing so.

 * Improve performance of \\afs\<cell><mount-type><volume>\
   access by eliminating extraneous attempts to search the 
   CellServDB file and DNS AFSDB records.  Share names 
   containing '%' or '#' are not valid cell names.

 * Return STATUS_RANGE_NOT_LOCKED if the application attempts
   to unlock a range that is not currently locked.

 * Enable the 32-bit tools to read the 64-bit registry keys
   when executing in the WOW64 environment.
   

Since 1.5.29
 * Fix buffer refcount leak introduced in 1.5.29

 * Prevent the NIM AFS Provider from crashing when configured to use 
   Kerberos v4 for token acquisition in situations where no Kerberos v4
   support is available.  For example, 64-bit KFW or sites that delete
   the Kerberos v4 DLL.

 * In afsd_init.log, support logging cache sizes greater than 2GB.

 * The volume status output of "fs examine" was being determined based
   upon the wrong error value.  Instead of using the pioctl() return
   code, the errno value must be used.  

 * Prevent invalid pts auto-registration attempts by aklog and the 
   NIM afs provider when Kerberos referrals are in use.  (krb5 1.6+)

 * Add synchronization protection to all of the SMB Listener State 
   variables.  This prevents race conditions when the addition or
   loss of a network adapter takes place.

 * On Vista, there is a race condition between the restart of the 
   afs client service and the network adapters.  Ensure that the 
   afs client service can handle switching between loopback and 
   non-loopback modes of operation.

 * Fix FollowBackupPath to work with mount points that are not 
   explicitly read/write.

 * Fix a deadlock that has been observed on Vista while resuming
   after at least four hours of sleep.

 * Add new HKLM registry configuration options that can be used 
   to provide configuration hints to the NIM AFS Provider when 
   creating new identities.  See release notes for details on
   HKLM\SOFTWARE\OpenAFS\Client\Realms\.

Since 1.5.28
 * Speed up write buffer operations by holding a global read-lock
   instead of a write-lock when updating the buffer data version 
   number.  The data version is protect by the associated stat
   cache mutex.

 * Speed up read buffer operations by removing out of date buffers
   from the buffer hash tables.

 * Protect against a null pointer dereference in afslogon.dll
   if QueryAdHomePathFromSid() fails.
   
 * Remove a race condition while accessing private freelance data
   structures.

 * Update the buffer data versions when Freelance directory data
   changes.

 * Complete the implementation of smb_ReceiveTran2GetDFSReferral().
   Critical for Vista.  

 * Return EAS_NOT_SUPPORTED in response to EA read/store/query 
   instead of returning zero length.

 * Return ERROR_RETRY when a server sends VBUSY.

 * Return PATH_NOT_COVERED instead of BADSHARENAME for DfsLinks when
   a volstat plug-in is in use.

 * Enhance the volstat interface to support Dfs path mappings.

 * Fix "symlink list" to do the right thing for Dfs links

 * Fix callback expiration that was broken in 1.5.27.

 * Fix a crash at shutdown caused by the lack of synchronization
   within the daemon threads (cm_Daemon, cm_BkgDaemon, cm_IPAddrDaemon)

 * Hold the cm_server_t mutex while walking or modifying the 
   cm_server_vols_t list in order to avoid a race condition.

 * Ensure that the cm_server_t refCount is safely modified.

 * The pthread library implementations of pthread_cond_wait
   and pthread_cond_timedwait were failing to return with 
   the mutex held when error conditions occurred.  This could
   result in race conditions wherever they are used.  One case
   in which problems were seen was the Rx event_handler().

 * Prevent cm_GetBuffer() from locking all buffers associated with 
   a stat cache object before deciding if the current buffer is up 
   to date.

 * Fix cm_BkgPrefetch to actually ensure that there is a cm_buf_t 
   object for each offset that we will attempt to reserve.

 * Reduce the default number of background daemon threads used to 
   prefetch or store to one in order to prevent two or more threads 
   from stepping on each other.

 * Reduce the default chunk size to 256KB in order to prevent read 
   timeouts on low bandwidth connections.

 * Fix a race condition that could result in threads becoming stranded 
   when synchronizing multiple requests on the same stat cache object.

Since 1.5.27
 * The GiveUpAllCallBacks functionality that was added in 1.5.21
   has been disabled.  It has been disabled because it turns out
   that the RPC is improperly implemented in all file servers 
   previous to 1.4.6.  The RPC is executed without the proper 
   locks being held resulting in data corruption and eventual file
   server failure.  The functionality will be re-enabled in a 
   a few months.

 * a couple of .readonly Volume Callback optimizations were 
   implemented:
   -  Apply most recent volume callback to all stat cache objects 
      in the volume.  This avoids premature callback expirations 
      and unnecessary FetchStatus calls.
   - Add option to permit attempts to renew .readonly Volume 
     Callbacks on a periodic basis to avoid invalidation of stat 
     cache data.  [Requires daemonCheckVolCBInterval be set to 
     non-zero number of seconds.  1800 (30 minutes) is suggested.]

 * Adds support for 64-bit data version values.  Previous releases 
   only supported 32-bit values.  Larger values sent by the file server 
   would result in rollover.

 * aklog and asetkey no longer generate an exception if Kerberos for 
   Windows is not installed.

 * If the LSA authentication functions report a temporary out of memory 
   condition, fail the current authentication and do not cause the AFS 
   client service to terminate unexpectedly.

 * When the AFS client service is shutting down, prevent the daemon check 
   thread from executing additional checks which might attempt to access 
   discarded locks or unmapped memory.

 * Correct a double free condition in the NIM AFS credentials provider.

 * Add "FollowBackupPath" service parameter registry value.   
   This provides equivalent functionality to the UNIX afsd -backuptree
   option.  When set to a non-zero value, normal mount points originating
   in a .backup volume will prefer a .backup volume.

Since 1.5.26
 * Fix NSIS uninstall from the .MSI installer.

 * Add Stack Frame logging for AMD64

 * Add support for EWOULDBLOCK to cm_Analyze.  If the file 
   server returns EWOULDBLOCK, retry the request every two 
   seconds for up to the RDRtimeout.

 * Reorganize the order of the includes to ensure that EWOULDBLOCK 
   is not assigned the same value as EIO.

 * Update CellServDB to GCO Public 25 Oct 2007

 * Add additional validation and error handling code after 
   each call to getSlot().  If an invalid slot is returned, 
   return NONODE.  If the invalid slot is returned when 
   extracting a data node, invalidate the tree.

 * Modify compareKeys() to always perform a case-insensitive 
   comparison and only perform a case sensistive comparison 
   if the case-insensitive one matches.  This ensures the 
   ordering is consistently reported.

 * Add lock assertions to ensure that all calls are being 
   performed with the correct locks being held.  There have been 
   some crash reports that provide stack data that does not appear 
   to be possible unless there is a race.  However, there are no 
   obvious locations where the race is taking place and the test 
   suite indicates that all of the correct locks are being held. 
   We shall see what happens in the field.

 * For consistency replace all calls to findKey in which the range i
   s (1,numentries) with calls to getSlot().

 * Optimize the depth search loop by testing the slot value in the 
   for statement instead of forcing the loop to be broken later.

 * Reorganize the locking for cm_BeginDirOp and cm_EndDirOp.
   There are a number of locations where locks are obtained, dropped, 
   and reobtained.  This reorganization attempts to accomplish several 
   things:

   (1) be optimistic for the most common case so it will be fast

   (2) add consistency checks after each location where locks are 
       dropped and re-obtained.  If we lose a race in cm_BeginDirOp 
       and the bplus tree is out of date, retry until we get to a 
       consistent state that we can use.

   (3) Ensure that all operations take place with the correct locks.

 * One of the issues that has become a serious problem since the 
   addition of the local directory updates is that although cm_SyncOp 
   synchronizes operations, it does not preserve the order of requests. 
   This has always been a problem in that it has been possible for a 
   request to fail to complete due to its worker thread's bad luck. 
   When a request takes longer than the Windows SMB Redirector's timeout, 
   the SMB Redirector tears down the SMB virtual circuit. 

   When using the local directory updates it is really important that 
   the directory update operations complete in the order that they were 
   sent to the file server. If they don't, then the local directory state 
   and the file server state will not match and the local directory state 
   must be discarded which in turn forces a new read of the entire 
   directory contents over the network. 

   This patch adds a new cm_scache_waiter_t object that is used to store 
   the current thread, buffer, and syncop flags within a waiters queue 
   on each cm_scache_t object. If a thread is forced to sleep in 
   cm_SyncOp, upon waking it will check to see if there are any other 
   threads waiting that are attempting to perform a similar task ahead 
   of it in the queue. If yes, the thread goes back to sleep. If not, 
   it goes ahead and enters the cm_SyncOp conflict resolution block. 

   This patch has the additional side effect of reducing the number of 
   competing threads that must obtain the cm_scache_t mutex and process 
   the cm_SyncOp conflict resolution block. As a result, the overall CPU 
   utilization of the service and the clock time associated with processing 
   requests will be reduced.

 * assert that the cm_scache_t mutex is held when calling
   buf_ForceDataVersion().

 * rename findNode to leafNode in bplus_Lookup

 * replace all OutputDebugString calls with osi_LogX calls

 * modify bestMatch to special case the return values for leaf nodes.
   If an entry is above or below the values available in the leaf node
   return BTLOWER or BTUPPER instead of BTERROR.

 * In insert and delete operations check for BTLOWER/BTUPPER and isleaf,
   if true convert to either slot 0 or Max and perform the insertion.
   This produces easier to read code when performing lookups.
 
 * modify lock_AssertXXX macros to call osi_assertx() and provide a
   descriptive message.

 * the difference between osi_assert() and osi_assertx() is whether or not
   a description is specified as part of the call.  When no message is
   specified afsd_notifier() is called with a NULL msgp parameter.  This
   results in a null pointer dereference during the EventLog() call.

   As a result, none of the information describing the assertion is written
   to the log file.

   This commit sets a default message for use when no other message is
   provided and it replaces all calls to osi_assert with osi_assertx and
   adds descriptions.
   
 * Migrate B+ tree search key into thread local storage

Since 1.5.25
 * Fix an uninitialized variable in lana_IsLoopback() which 
   could result in a random determination that a loopback 
   adapter is not a loopback adapter.  This would result in 
   a computed Netbios service name of %MACHINE%-AFS instead
   of AFS.  This is a serious problem if the value computed
   by the AFS Client Service and the client process pioctl
   calls produce different results as it will result in a  
   communication failure.

 * Fix a crash in the B+ Tree code that results when a file
   rename only results in a change of case.

 * Fix the pioctl path parsing.  The previous algorithm 
   broke the current directory of the client into two parts,
   the TIDpath which represented the device root and the
   relative path.  The problem is that this is not sufficient
   from proper processing of symlinks and mount points.  

   What is required is that the last component of the path
   must not have the symlink or mount point be evaluated
   whereas symlink and mount point evaluation must be performed
   for the rest of the relative path.  Therefore, the path is
   now split into three parts: TIDpath, intermediate path,
   and last component.  

   "fs examine" now reports the data of the symlink or mount
   point object instead of the target and "symlink list" and
   "symlink remove" now works.  The "Explorer Shell Extension"
   menu now is displayed for symlink objects.

 * More informative symlink.exe error messages.

 * When the cmd.exe "del" command is used to remove a symlink,
   not only was the link object removed, but the stat cache
   object for the target was marked deleted.  This resulted
   in cache corruption which would require the destruction
   of the cache to correct.

 * Add undocumented registry value, "OfflineReadOnlyIsValid"
   which results in cached contents from read only volumes
   to be treated as valid when there is no callback registered
   and all servers are down or all instances are offline.

   HKLM\SYSTEM\CurrentControlSet\Services\TransarcAFSDaemon\Parameters

 * Add documented registry value, "AcceptDottedPrincipalNames"
   which is used to permit the acquisition of AFS tokens 
   using Kerberos v5 principal names which include a dot in
   the first component.

   {HKLM\HKCU}\Software\OpenAFS\Client

 * "fs uuid" does not require "AFS Client Administrator" group
   membership to display the current UUID value.

Since 1.5.24
 * Update NSIS installer to version 2.30

 * Update Wix installer to version 2.0.5325
 
 * 64-bit KFW support:
   - Network Identity Manager plug-in
   - 64-bit aklog
   - 64-bit krb5 enable afscreds.exe, afs server manager,
     etc.

 * Enable B+ tree Directory Search by default.

 * NIM plug-in is now compatible with FILE: ccaches

 * Install instloop.exe into Client/Programs/ even though
   it isn't required.   Permit it to be used by organizations
   that clone Windows systems.

 * On a suspend, only drop callbacks for servers that 
   are online.  If the server is offline, maintain the callback
   until the server is once again available.

 * Prevent against acquisition of a directory write lock when
   there are no updates to be performed.  This was resulting 
   in a write-lock leak which later on would lead to a deadlock.

 * Add configurable file pre-fetch mode.  For files that match
   the specified file extensions, pre-fetch the entire file into
   the cache.

   HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
      (MULTI_SZ) PrefetchExecutableExtensions

 * Fix the client_osi large integer support so that return values
   can be properly assigned to large integer variables.
 
 * Do not hold cm_scache_t mutex across cm_GetCell() calls 
   since cm_GetCell() can block in a RPC.

 * Add short file name support to B+ tree searches.

 * Resolve a conflict between the desire to notify file servers
   that the client workstation is suspending and the desire to
   permit objects with valid callbacks when the client suspends
   to continue to use those objects from the cache if the client
   resumes without network connectivity.

 * Fix stack variables to ensure that all buffers storing paths
   can handle AFSPATHMAX (1024) characters.  Paths of 256 characters
   would cause a stack overflow condition.

 * On Vista, force afs_config.exe to run with the highest privileges
   available to the end user.  For a user in the Administrators
   group running under UAC, this will cause afs_config.exe to be 
   be run with full Administrators privileges after the user 
   approves the privilege escalation.

 * More VNOVNODE handling issues.  If a dirty buffer cannot be 
   written due to VNOVNODE, clear the buffer and ignore the error.
   If a cm_scache_t object has been marked deleted due to a VNOVNODE
   error, the next time the smb_fid_t object that references it is 
   used return either NOSUCHFILE or NOSUCHPATH and close the
   smb_fid_t object.  Subsequent attempts to use the smb_fid_t will
   result in an invalid handle error.

 * For Vista, set the RequestExecutionLevel of the NSIS installer
   to "admin".

 * For the build environment, support the generation of manifests
   when using the CL1310 (VS .NET 2003) compiler/linker.

 * There was a bug in OutputDebugHexString() which could overflow
   a stack buffer resulting in unexpected process termination
   of afsd_service.exe if the input Security Token contained
   many bytes equivalent to the '%' char.

 * Add directory lookup statistics to "fs memdump" output.
 

Since 1.5.23
 * Windows uses case-insensitive file name pattern matching
   but AFS is a case sensitive file system.  The AFS3 directory
   format is block based, uses network byte order and 
   includes a hash table for fast case sensitive lookups.  
   This causes several problems for the Windows AFS client.
   (1) Traversing the directory blocks is cpu expensive
   (2) A hash table miss does not indicate that the desired
       entry does not exist.
   (3) Determining whether a non-ambiguous inexact match or
       the entry does not exist requires a linear traversal
       of the entire directory.  
   These issues often result in 100% CPU utilization.

   These issues are addressed by building a modified B+ tree for 
   each directory and then using the B+ tree for searches.

   This feature is disabled by default.  To enable it set
   the "BPlusTrees" registry value to 0x01 in the AFS Client
   Service Parameters key.

 * Correct a null pointer dereference in the symlink recursion
   detection code.  (Introduced in 1.5.22)

 * Add the Network Identity Manager AFS Provider to AMD64 
   builds.

Since 1.5.22

 * A new registry value, "BlockSize", can now be used to adjust 
   the size of the internal buffers used to store file data.  
   The default is 4KB.  When set, the values 1..1024 specify
   multiples of 4KB. Values > 1024 specify actual block sizes
   and must be multiples of 4KB.  The actual block size in KB 
   can not be larger than the chunk size which defaults to 1MB.

    HKLM\SYSTEM\CurrentControlSet\Services\TransarcAfsDaemon\Parameters
      (DWORD) BlockSize

 * pts.exe and ptserver.exe built with -DSUPERGROUPS

 * Service Manager shutdown hints have been added.  The hint 
   specifies that it may take the AFS client service up to two
   minutes to shutdown. 

 * Fix the recursion detection feature introduced in 1.5.22.
   The previous implementation broke the evaluation of paths
   using ../path or ./path.   It also broke the evaluation of
   symlinks to absolute paths.  For example, /afs/cell/path.

 * Fix an uninitialized variable in cm_MakeDir that could result
   in crash if the requested directory could not be created.

 * Fix a the VolumeStatusNotification log messages to save
   the input string in the circular log buffer so garbarge 
   won't be output to afsd.log when the log is dumped.
   

 * Increase the SMB Maximum Path length to 260 characters
   from 256 as that is what CIFS supports.

 * On 64-bit Windows, use a unique description string for 
   the 32-bit AFS Shell Extension.

Since 1.5.21

 * In the AFS Admin Library, replaced all of the 
   Ubik_Call(XXX_RPC,...) calls with Ubik_XXX_RPC(...).  This 
   allows parameter checking and proper type conversions to be
   performed.  This is crucial when passing parameters that are
   64-bit wide on systems in which 'long' is a 32-bit value.
   Otherwise, the parameter list on the stack is interpretted
   incorrectly and the function not only doesn't succeed but it
   might corrupt memory and if lucky crash the application.
  
   The only program that we care about that uses this library 
   is the AFS Server Manager.  Now it doesn't crash.

 * Fixed smb directory searches for non-wildcard names.  The 
   optimized function should not have set the bulk in progress
   flag.  This would adversely affect later directory searches
   that required a bulk search.

 * Detect infinite recursions (up to 512 stat cache objects
   in a chain.)  If an infinite recursion is detected, return
   an error to stop the CIFS client from querying us forever.

 * Executables executed out of AFS name space can have their
   pages swapped out by Windows.  Those pages will be reloaded
   when they are required.  However, if the cache manager no
   longer has a callback and all of the file servers with 
   the volume data are down, then the application will crash
   due to the inability to read the required page.  

   We now track the down time of each server.  If the callback
   expires after the server that issued it went down and all
   servers are down, we delay the expiration to permit the data
   in the cache to be continue to be used by Windows or an 
   application.

 * The Windows AFS client has always had very poor performance
   with regards to directory management.   In the case of 
   directory updates caused by create, rename,and delete operations,
   the AFS cache manager will now perform local directory updates
   when possible in order to avoid re-reading the directory from
   the file server.  This will be done in whenever both of the 
   following conditions are true:

   - the data version on the directory increased by one as a 
     result of the operation

   - all of the directory buffers necessary to make the change
     are present in the cache

 * The error message in the AFS Control Panel describing the 
   need to be a Windows Administrator has been replaced with
   a message describing the need to be in the AFS Admins Group.

 * If the vldb server list is empty when attempting to evaluate
   a cell mountpoint it was possible for a crash to occur.  This
   can only happen if there is an existing mount point and the
   is a failure attempting to load the server list from the 
   CellServDB file or via AFSDB queries.

 * dirty buffers only have their dirty bytes stored to the file
   server instead of the entire chunksize

 * the default chunksize has been increased from 128KB to 1MB
   

Since 1.5.20

 * changed the enum values for cm_serverRef_t state info to use a 
   private name space to avoid collisions (srv_)
 
 * added a srv_deleted state for cm_serverRef_t objects.  This 
   state is set when cm_FreeServerList() is called with the 
   CM_FREESERVERLIST_DELETE flag set.  cm_FreeServerList() may
   not always delete the cm_serverRef_t from the list if it is
   still in use by another thread.  the srv_deleted state means
   the object's contents are no longer valid and it must be 
   skipped.  It will be deleted the next time the object is 
   freed and the refcount hits zero.
 
 * the srv_deleted state is also used when a file server reports
   either VNOVOL or VMOVED instead of marking the cm_serverRef_t
   as offline.  This is done to prevent additional usage of the 
   stale vldb data while waiting for the update volume request
   to complete.
 
 * added a state field to the cm_volume_t object (enum volstate 
   vl_ name space) that maintains the state of the volume based
   upon the states of all of the cm_serverRef_t and cm_server_t
   objects.
 
 * modified cm_UpdateVolume() to set the state of the cm_volume_t
   RW, RO, and BK to either vl_alldown or vl_online.  There can't
   be any other states because cm_UpdateVolume() destroys any  
   previous knowledge we might have had regarding busy or offline
   volume status
 
 * modified cm_UpdateVolume() to update the volume name in the 
   cm_volume_t to the volume base name if the previous value was 
   a volume ID.
 
 * modified cm_FollowMountPoint() to check to see if the volume
   name is a volume ID and if so call cm_GetVolumeByID instead 
   of cm_GetVolumeByName.  This ensures that volume IDs are always
   looked up as numeric values.  There is no longer a need to 
   maintain a separate cm_volume_t containing the string representation
   of the ID value.
 
 * Added a flags parameter to cm_GetVolumeByName() and cm_GetVolumeByID().
   The first flag is a "CREATE" flag which is set by all existing
   calls.  The flag is not set by calls to cm_GetVolumeByID() from
   the server probe code when volume status is being updated.  We 
   do not want the server probe operation to result in additional 
   turnover in the cached volume data.  The second flag is NO_LRU_UPDATE 
   which is set when the server probe code updates the volume status.
   This flag will be used to prevent the server probe operation from
   changing the order of the least recently used queue.
 
 * Modified cm_GetVolumeByName to ensure that only one cm_volume_t is 
   allocated for a given set of normal, readonly, and backup volumes
   regardless of whether or not the volume is accessed via name or 
   ID number.  The cm_volume_t namep field is always the base name
   of the volume. 
 
 * Added a new volume state, vl_unknown.  This state is used as
   the initial state for all cm_volume_t when the cache manager starts, 
   for each cm_volume_t at creation, and for each cm_volume_t when
   recycling.  The cache manager does not know the state of all 
   volumes in the world, only those that are in the cache and for
   which it has queried the VLDB and hosting file servers.
 
 * modified cm_GetVolumeByName() to initialize the state of a 
   volume to vl_unknown.  The actual state will be set when a 
   cm_VolumeUpdate() call completes successfully.
 
 * changed name of scache hash table variables to avoid ambiguity
   when adding hash tables for volumes
 
 * fix a buffer overrun in sys\pioctl_nt.c  pioctl().
 
 * modified cm_UpdateVolume() to handle the case in which there is
   no RW volume but there is are RO volumes for a given base name. 
   This is done by querying for the ".readonly" volume name if the 
   base name does not exist in the VLDB.  We never query for the
   .backup name because under the current usage model a .backup 
   volume may only exist on the server that the read-write volume
   is located.  If there is no RW volume, there can be no .backup.
 
 * Added four hash tables for cm_volume_t objects to improve the 
   search time of cm_GetVolumeByID(), cm_GetVolumeByName() and
   cm_ForceUpdateVolume(). One each for Name, RWID, ROID, and 
   BKID.  Three ID hash tables are necessary as long as it is
   desireable to maintain a single cm_volume_t containing all
   of the related RW, RO, and BK volume data.  Having the RW and
   RO volume data in the same object is necessary for the 
   implementation of cm_GetROVolumeID() which returns either the
   RO or RW ID depending upon the existence of RO volume instances.  
 
 * Added a volume LRU queue so that volume reuse becomes fairer.
   This does not replace the all Volumes list which is used when
   it is desireable to walk a list of all the volumes whose order
   is not going to change out from underneath you which makes it
   safe to drop the cm_volumeLock.
 
 * handles volume hash table updates where volume name to 
   volume ID number changes.  The volume name remains 
   constant in the cm_volume_t.  if a vos rename is performed, 
   the name of the volume will change and the volume IDs will be 
   updated.  Subsequent access to the old volume ID will create a 
   new cm_volume_t with the new name.
 
 * Added a daemon thread operation to query the state of volumes
   listed as busy or offline.  cm_CheckBusyVolumes() calls 
   RXAFS_GetVolumeStatus() for each volume ID that is marked vl_busy 
   or vl_offline.   If the volume is now online, the status on the
   volume is updated.  The default period is 600 seconds.  This can 
   be configured with the BusyVolumeCheckInterval registry value.  
 
 * Added prototype for smb_IoctlPrepareRead() which was missing a
   return type in the function definition.
 
 * Added volume id lists to the cm_server_t.  These lists are
   allocated in blocks of ~32 IDs.  When a cm_PingServer() 
   detects a change in server state, the state of the cm_volume_t
   is updated.  
 
 * Added volID to the cm_serverRef_t object.  volID is used
   to identify the volume for which the object is a referral.
   cm_FreeServerList() uses the volID to remove the volume 
   from the cm_server_t.
 
 * In cm_Analyze, when VNOVOL or VMOVED are received, 
   call cm_ForceVolumeUpdate() to force a refresh of the volume 
   location data.
 
 * Added cm_VolumeStatusNotification() which is used at the moment
   to log volume status changes to the trace log.  It will also
   be used as the access point to the File System Filter driver
   notification engine.
 
 * Added an all cm_scache_t list to cm_data.  This replaces the use
   of the stat cache LRU queue when we need to enumerate all 
   entries.  The LRU list order is not static and when using it to 
   enumerate all entries it can result in items being missed or
   items being processed more than once.
 
 * Modified cm_Analyze(). Instead of reseting the busy or offline
   state of a volume and forcing a retry of the operation
   cm_Analyze will defer to the background daemon thread that will
   update the state once every 600 seconds.
 
 * Added the automatic generation of a Freelance ".root" read-write
   mountpoint that refers to the root.afs volume of the workstation
   cellname at the time the mountpoint is created.
 
 * re-write cm_Analyze to make better use of the known volume
   status.  VL_Server queries cannot result in CM_ERROR_ALLOFFLINE
   messages.
 
 * renamed cm_CheckBusyVolumes to cm_CheckOfflineVolumes.  
   busy volumes will be reset to srv_non_busy by the function
   but there is no mechanism for querying the busy state other
   than by attempting to access the resource.
 
 * cm_Analyze will query the state of an offline volume before
   deciding whether or not to retry when all volume instances
   are offline.
 
 * the algorithm used after an IP address list change detection
   has been modified.  we now check the state of the VLDB servers,
   then reset the rx connections, and finally check the state of 
   the file servers.  Previously we flushed the rx connections 
   and then checked the servers in whatever random order they 
   were first discovered.
 
 * fix the volume status reported by "fs examine"
 
 * various improvements to the AFS Network Identity Manager Provider.
   - Don't access freed objects after the window has been destroyed.
   - Write back the AFS cell list in the identity configuration 
     even when its empty.
   - Disable all use of the provider when AFS is disabled for a 
     particular identity.
 
 * Fix cm_cell_t leak.  When evaluating a mountpoint that uses an
   alias instead of the full cellname, double check the full cell
   name and see if we have a matching cell before allocating a new
   one.
 
 * Dump more data with "fs memdump". All volumes and scaches and
   now cells.
 
 * Reverse the order of the all cells list.  This ensures that
   the workstation cell is always first in the list.
 
 * If we recognize the server from which a callback revoke was
   received, then we can ensure that we only invalidate the status
   information for cm_scache_t objects belonging to that cell.
 
 * When building for amd64, use APPVER 5.02 as we do not support
   64-bit Windows 2000 and explicitly state the machine type in
   the link flags.  
 
 * no longer treat VMWare network adapters as equivalent to a 
   loopback adapter.  For some unknown reason, publishing the "AFS"
   netbios name on them frequently results in a name collision 
   error.
 
 * Use read locks whenever possible instead of write locks when
   walking the allCellsp list.  This permits better use of multiple
   CPUs.
 
 * In 1.5.20 a bug was introduced by the fake status when PRSFS_READ
   is not present on the directory optimization.   It would report
   mount points and directory objects for which there was no
   status information as a zero length file.  Instead we can use 
   the vnode number of the object to determine if it should be
   considered to be a directory or a file.  If it is odd it is a
   directory; even it is a file.
 
 * Give Back Callbacks to the file servers in response to suspend
   and shutdown power management events.  Only give up callbacks
   to UP servers and do so with a very short RX dead timeout.
   (10 seconds)
 
 * Fix a race condition in the SMB FID cleanup code that could 
   result in a stat cache mutex never being released.
 
 * Fix the graphics files that are installed as part of the 
   IBM Administrators Guide.
 
 * Change the behavior of pioctl path parsing to not follow 
   symlinks or mount points during evaluation.  This results in
   the FID being the symlink or mount point and not the object
   they point to.  This is consistent with the behavior of the 
   UNIX client.

 * Enable AFSDB record lookup for command line tools such as
   "fs" and "vos".

 * CellServDB update 2007-07-05

 * Added a Volume Status Plug-in Interface that permits third parties
   to receive notifications when the volume state changes occur.

 * Ensure that all writes to the afsd_alloc.log file use CR-LF.

 * Prevent the afsprocmgmt.dll from being unloaded while the 
   background thread it starts is running.

 * Avoid a race condition when closing smb_fid_t objects.

Since 1.5.19
 * fix install path of IBM Quick Start Guide when installed by
   the Wix installer

 * fix the client so that it can write into directories with
   only 'li' permissions.

 * fix installation of 32-bit Shell Extension for 64-bit systems.
   Install it under its own HKCR keys.

 * modify the wix installer to remove and upgrade install packages
   with the same version number.  

 * when flushing volume location data every two hours or in 
   response to "fs checkvolumes", also flush the mount point 
   links in order to force their re-evaluation.

 * aklog was copying an uninitialized buffer as part of the 
   foreign cell automatic pts registration

 * when obtaining directory listings, check for read permission
   on the directory.  If read permission does not exist, fake the
   status information for the directory entries.  This improves
   performance, prevents the client from pounding the file server,
   and prevents leakage of cached status information.

 * "fs getcacheparms" output was incorrect.

 * NetIdMgr AFS Provider should only automatically add token 
   acquisition for cell foo.com if the realm of the cell matches
   the realm of the identity.

 * Fix a null pointer dereference in cm_MergeStatus()

Since 1.5.18

 * network identity manager provider:

   - Correct for the fact that pr_CreateUser tampers with the 
     contents of the username input buffer.

   - Take into account that the krb4 credential data structure
     does not have separate realm fields for the user and service.
     When krb524 is used with a cross-realm ticket, the realm
     describes the service not the user.  

 * Many integrated logon changes:

   - Fix eventlog reporting.  Do not attempt to log an event 
     if the event source registration fails.  Use DebugEvent0 
     instead of DebugEvent when there are no parameters.

   - Modify the LOOKUPKEYCHAIN macro to recognize ERROR_MORE_DATA 
     errors.

   - Fix the reading of Domain specific configuration for 
     LogonScript and TheseCells.  Previously the dwSize value was 
     being overwritten so that subsequent RegQueryValueEx call 
     would fail.

   - Fix a memory leak in the TheseCells reading code.

   - Add support for Domain specific "Realm" specification.  The 
     realm is the realm to be appended to the username.  When 
     logging in as a domain or to the local machine, the specified 
     "Domain" name is not going to be a valid realm name.

   - Construct a proper principal name based upon the domain 
     specified realm for use in obtaining tokens with KFW.

   - If the domain specified "TheseCells" list includes the 
     default cell, do not obtain tokens twice.

 * An unprotected call to the Kerberos v5 function, 
   krb5_free_context(), can result in an invalid memory access
   exception if the contents of the krb5 profile, krb5.ini, cannot
   be parsed correctly.  The error is located within the afskfw
   library which is used by afscreds.exe, afssvrmgr.exe, and 
   afslogon.dll.  When the error is triggered within afslogon.dll
   a blue screen condition can occur when the winlogon.exe system
   process terminates unexpectedly.

Since 1.5.17

 * Do not stop the afs client service with an assert
   if the cell name cannot be determined when responding
   to a "fs whereis" command.

 * Integrated Logon (afslogon.dll).  No longer attempt 
   to preserve Kerberos 5 tickets for the logon session
   as that job is now supported by KFW 3.1 and above.

 * The Network Identity Manager plug-in will attempt to
   remove expired service tickets from the cache.

Since 1.5.16

 * CellServDB update - GCO Public CellServDB 09 Mar 2007

 * AFS Server Manager and AFS Account Manager data cache
   moved from HKCU\Software\TransarcCorporation to 
   HKCU\Software\OpenAFS to prevent crashes due to data
   format changes.

 * Reduce the default UP server probe time from 10 minutes
   to 5 minutes to match UNIX.  This will not solve the NAT
   port mapping timeout issue because most NATs default to
   a value less that 5 minutes and this value ensures that
   the afs client server will start probes 5 minutes after
   the last probe set ended.  The probe period for any one
   server is therefore longer than 5 minutes.  The more
   servers in the list the longer the time between probes.

 * Before marking a server as down in cm_Analyze, retry the
   request using a new rx connection to the server.  If the
   cache manager is behind a NAT, it is possible that the 
   response from the server is being sent to the original   
   address:port from which the rx connection was established.
   This will appear to the cache manager as the server being
   down when in fact it is not.  Forcing the use of a new rx  
   connection will result in the server responding to the    
   most recent address:port assigned to the client by the NAT.

   If the server is in fact down, it will simply take a little
   longer to notice it.   

Since 1.5.15

 * Removing a directory that contained entries was failing with
   the wrong error code being returned to the CIFS client.

 * rxdebug -long 

 * rxdebug exponential back-off.  don't take down a server that
   is already struggling to respond.

 * Network Identity Manager plug-in registers help on the Help
   menu if it is supported by NIM.  (KFW 3.2 and above)

 * Plug-n-play networking did not always work after coming back
   from a Windows Suspend.  The AFS client service failed to 
   re-compute the LAN adapter to bind to and instead attempted
   to bind to all LAN adapters.  This would polute the public
   network with the name "AFS" which could lead to conflicts 
   between machines.

 * Build installers using Microsoft Hotfixes provided at
   http://tinyurl.com/243svk for Visual Studio C RunTime Library
   Daylight Savings Time issues.  

 * "fs checkserver" does not require admin privileges unless 
   the probe interval is being changed.

Since 1.5.14

 * A reference leak was discovered in the mountpoint traversal 
   functionality which resulted in cm_volume_t objects not becoming
   eligible for re-use.  This will trigger a panic condition when
   more than the maximum number of volumes is accessed.

 * Fixed the success test for the "fs memdump" operation.  Success
   is now reported as such.  Added cm_volume_t object information 
   to the dump file.

 * A data corruption error has been fixed.  Write requests
   queued for background processing were not retried upon failure.
   This leads to inconsistencies between the contents of the file
   in the local cache and the contents of the file on the file 
   server.  

   The implemented fix is to re-queue any request that has failed
   due to timeout, volume busy, volume offline, server busy, or
   server offline.  Requests are not processed while all servers
   for a volume are inaccessible.  When servers become accessible
   the background request will be retried.

   In addition, if a background write fails, do not mark the buffers
   as clean.  Failure to preserve the dirty flag prevents the buffers
   from being written to the file server in the future.

   This fix does not address the possibility that the credentials
   necessary to perform a write request might expire between the
   time the request is queued and when the request can be processed
   due to server accessibility.

 * The Freelance volume does not support per user ACLs.  The
   cm_HaveAccessRights function did not handle this special case.
   Requests for privileges greater than read|lookup would result
   in afsd_service.exe entering an infinite loop.  ISOBuster is
   known to cause this behavior.

 * When deleting files, the SMB Server would attempt to obtain
   the access permissions for the parent directory instead of 
   for the file itself.  This would result in an incorrect 
   assessment of whether or not the file can be deleted.

 * The Network Identity Manager AFS plug-in has had its string
   tables updated for consistency with the language used by NIM.
   All credentials are referred to as credentials instead of tokens.
   Drop-down combo boxes now in fact drop down.

 * The AFS Salvager was not properly being built on Windows.
   This is relevant only if the AFS servers are in use.

 * The default signal handlers for the AFS servers incorrectly
   processed SIGQUIT.  Instead of stopping the server, an
   exception was raised.

 * Fix another volume refcount leak.  This one was triggered by
   an inability to contact the vlservers in order to update the 
   file server list.

 * Fix checklist control used by the drive mappings tabs in 
   afscreds.exe and afs_control.exe.

 * Improve output of "fs memdump" related to buffer management.

 * Fix refcount leak of cm_buf_t objects when prefetching buffers
   already in the process of being fetched.

 * Improve cache performance by computing the buffer hash table
   size based upon the number of buffers so that the average 
   number of entries in a bucket is seven.

 * Improve cache performance of read/write data by locally 
   updating the dataVersion of all buffers associated with the
   file on each StoreData or StoreStatus operation provided that
   the dataVersion was incremented by one and the dataVersion of
   the buffer matched the original dataVersion of the cached 
   status information for the file.

 * When recycling cm_scache_t objects, make sure that we 
   clear all of the previous fields.  It is no longer the
   case that we can assume that the stat cache object is being
   obtained only after a successful FetchStatus call.  
   InlineBulkStatus will allocate stat cache objects for which
   the no status information is available.

 * When a file is being opened with the OPEN_ALWAYS disposition,
   do not require write permissions when the file already exists
   unless of course the file is being opened requesting write 
   access.

 * When a file has been opened for read-only access, if the 
   application requests an exclusive lock, obtain a read-lock
   instead of a write-lock.  The purpose of the exclusive lock
   in this situation is to prevent changes to the file and the
   full file read lock will provide this guarrantee.


Since 1.5.13

 * Enforce Unix Mode Write bit to prevent file deletion of read only
   files.

 * The CIFS READ and LIST permissions use the same value 0x0001.
   When a directory is being accessed, check for PRSFS_LOOKUP 
   instead of PRSFS_READ.

Since 1.5.12

 * Fix computations of short names of mount points.  They were
   broken as part of the directory search optimization. (1.5.13a)

 * When all buffers are in use, periodically sleep the threads
   waiting for a buffer to become free while the global buffer
   lock is not being held in order to permit those threads that
   can free buffers to do so.

 * Power Management improvements.  Maintain a global flag that 
   specifies whether or not the service is in a suspend state.
   Do not panic if Netbios() returns NRC_BRIDGE meaning that the
   lana is no longer valid.  Instead, stop the listener threads
   and if all listener threads are stopped, reset the lana_list.
   Allow the cm_Daemon() thread to periodically check the state
   of the smb listeners.  If they are all stopped and the service
   is not suspended, attempt to restart them.  If there are no valid
   lanas, return to the stopped state.

 * CreateX and NTCreateX use cm_CheckNTOpen() to test whether or not
   the user is permitted to obtain read or write locks.  This function
   would obtain the lock and then drop it returning whether or not