Knowledge Base ISC Main Website Ask a Question/Contact ISC
DHCP 4.3.2rc2 Release Notes
Author: Michael McNally Reference Number: AA-01253 Views: 5561 Created: 2015-02-25 15:15 Last Updated: 2015-02-25 15:15 0 Rating/ Voters
          Internet Systems Consortium DHCP Distribution
                 Version 4.3.2rc2
                 26 February 2015

                  Release Notes

                  NEW FEATURES

The major "theme" for ISC DHCP 4.3.x was to update the support for
DHCPv6 to include several of the features that have been available
for DHCPv4.  These include:

- Support the use of classes

- Support for on_commit, on_expiry and on_release statements

- Better logging of address assignments

- Support for using DHCPv6 relay options in expressions

This release also adds suppport for the standard DDNS as described in the
current RFCs as well as enhancing support for dynamically adding and removing
subclasses via OMAPI.

There are a number of DHCPv6 limitations and features missing in this
release, which will be addressed in the future:

- Only Solaris, Linux, FreeBSD, NetBSD, and OpenBSD are supported.

- DHCPv6 includes human-readable text in status code messages, in
  English.  A method to reconfigure or support other languages would
  be preferable.

- The "host-identifier" option is limited to a simple token.

- The client and server can only operate DHCPv4 or DHCPv6 at a time,
  not both.  To use both protocols simultaneously, two instances of the
  relevant daemon are required, one with the '-6' command line option.

For information on how to install, configure and run this software, as
well as how to find documentation and report bugs, please consult the
README file.

ISC DHCP uses standard GNU configure for installation. Please review the
output of "./configure --help" to see what options are available.

The system has only been tested on Linux, FreeBSD, and Solaris, and may not
work on other platforms. Please report any problems and suggested fixes to

ISC DHCP is open source software maintained by Internet Systems
Consortium.  This product includes cryptographic software written
by Eric Young (

            Changes since 4.3.2c1

- Corrected a compilation error introduced by the fix for ISC-Bugs #37415.
  The error occurs on Linux variants that do not support VLAN tag information
  in packet auxiliary data.  The configure script now only enables inclusion
  of the VLAN tag-based logic if it is supported by the underlying OS.
  [ISC-Bugs #38677]

            Changes since 4.3.2b1

- Specifying the option, --disable-debug, on the configure script command line
  now disables debug features.  Prior to this, specifying --disable-debug
  incorrectly enabled debug features. Thanks to Gustavo Zacarias for reporting
  the issue.
  [ISC-Bugs #37780]

- Unit test execution now uses a path augmented during configuration
  processing of the --with-atf option to locate ATF runtime tools, atf-run
  and atf-report. For most installations of ATF, this should alleviate the
  need to manually include them in the PATH, as was formerly required.
  If the configure script cannot locate the tools it will emit a warning,
  informing the user that the tools must be in the PATH when running unit
  Secondly, please note that "make check" will now exit with a failure status
  code (non-zero) if one or more unit tests fail.  This means that invoking
  "make check" from an upper level directory will cause the make process to
  STOP after the first test subdirectory with failed test(s).  To force all
  tests in all subdirectories to run, regardless of individual test outcome,
  use the command "make -k check".
  [ISC-Bugs #38619]

            Changes since 4.3.1

- Corrected parser's right brace matching when a statement contains an error.
  [ISC-Bugs #36021]

- TSIG-authenticated dynamic DNS updates now support the use of these
  additional algorithms: hmac-sha1, hmac-sha224, hmac-sha256, hmac-sha384,
  and hmac-sha512
  [ISC-Bugs #36947]

- Added check for invalid failover message type. Thanks to Tobias Stoeckmann
  working with the OpenBSD project who spotted the issue and provided the
  [ISC-Bugs #36653]

- Corrected rate limiting checks for bad packet logging.  Thanks to Tobias
  Stoeckmann working with the OpenBSD project who spotted the issue and
  provided the patch.
  [ISC-Bugs #36897]

- Log statements depicting what files will be used by the server now occur
  after the configuration file has been processed.
  [ISC-Bugs #36671]

- Addressed Coverity issues reported as of 07-31-2014:
  [ISC-Bugs #36712] Corrects Coverity reported "high" impact issues.
  [ISC-Bugs #36933] Corrects Coverity reported "medium" impact issues
  [ISC-Bugs #37708] Fixes compilation error in dst_api.c seen in older
  compilers that was introduced by #36712

- Server now supports a failover split value of 256.
  [ISC-Bugs] #36664]

- Remove unneeded error #defines.  These defines were included in case
  external programs required the older versions of the macro.  They
  have been #ifdeffed for now and will be removed at a future date.
  See site.h for the #define to include them again, but you should
  switch to using the DHCP_R_* versions instead of the ISC_R_* versions.
  Also ISC_R_MULTIPLE has been removed as it is also deifned in bind.
  [ISC-Bugs #37128]

- Added checks in range6 and prefix6 statement parsing to ensure addresses
  are within the declared subnet. Thanks to Jiri Popelka at Red Hat for the
  bug report and patch.
  [ISC-Bugs #32453]
  [ISC-Bugs #17766]
  [ISC-Bugs #18510]
  [ISC-Bugs #23698]
  [ISC-Bugs #28883]

- Addressed checksum issues:
  Added checksum readiness check to Linux packet filtering which eliminates
  invalid packet drops due to checksum errors when checksum offloading is
  in use.  Based on dhcp-4.2.2-xen-checksum.patch made to the Fedora project.
  [ISC-Bugs #22806]
  [ISC-Bugs #15902]
  [ISC-Bugs #17739]
  [ISC-Bugs #18010]
  [ISC-Bugs #22556]
  [ISC-Bugs #29769]
  Inbound packets with UDP checksums of 0xffff now validate correctly rather
  than being dropped.
  [ISC-Bugs #24216]
  [ISC-Bugs #25587]

- Added the echo-client-id configuration parameter to the server configuration.
  The server now supports RFC 6842 compliant behavior by setting a new
  configuration parameter, echo-client-id.  When enabled, the server will
  include the client identifier option (Option code 61) if received, in its
  responses.  The server identifier returned in NAKs (if enabled) will now
  be the globally defined value (if one) if the server cannot attribute the
  inbound request to a known subnet.
  [ISC-Bugs #35958]
  [ISC-Bugs #32545]

- Added support of the configuration parameter, use-host-decl-names, to
  BOOTP request handling.
  [ISC-Bugs #36233]

- Added logic to ignore the signal, SIGPIPE, which ensures write failures
  will be delivered as errors rather than as SIGPIPE signals on all OSs.
  Thanks to Marius Tomaschewski from SUSE who reported the issue and provided
  the patch upon which the fix is based.
  [ISC-Bugs #32222]

- In the failover code, handle the case of communications being interrupted
  when the servers are dealing with POTENTIAL-CONFLICT.  This patch allows
  the primary to accept the secondary moving from POTENTIAL-CONFLICT to
  RESOLUTION-INTERRUPTED as well as handling the bind update process better.
  In addition the code to resend update or update all requests has been
  modified to send requests more often.
  [ISC-Bugs #36810]
  [ISC-Bugs #20352]

- By default, the server will now choose the value to use in the forward DNS
  name from the following in order of preference:

    1. FQDN option if provided by the client
    2. Host name option if provided by the client
    3. Configured option host-name if defined

  As before, this may be overridden by defining ddns-hostname to the desired
  value (or expression).  In addition, the server logic has been extended to
  use the value of the host name declaration if use-host-decl-names is enabled
  and no other value is available.
  [ISC-Bugs #21323]

- DNS updates were being attempted when dhcp-cache-threshold enabled the use of
  the existing lease and the forward DNS name had not changed.  This has been
  [ISC-Bugs #37368]
  [ISC-Bugs #38686]

- Corrected an issue which caused dhclient to incorrectly form the result when
  prepending or appending to the IPv4 domain-search option, received from the
  server, when either of the values being combined contain compressed
  [ISC-Bugs #20558]

- Added the server-id-check parameter to the server configuration.
  This parameter allows run-time control over whether or not a server,
  participating in failover, verifies the dhcp-server-identifier option in
  DHCP REQUESTs against the server’s id before processing the request.
  Formerly, enabling this behavior was done at compilation time through
  the use of the #define, SERVER_ID_CHECK, which has been removed from site.h
  The functionality is now only available through the new runtime paramater.
  [ISC-Bugs #37551]

- During startup, when the server encounters a lease whose binding state is
  FTS_BACKUP but whose pool has no configured failover peer, it will reset the
  lease's binding state to FTS_FREE.  This allows the leases to be reclaimed
  by the server after a pool's configuration has changed from failover to
  standalone. Prior to this such leases would remain stuck in the backup state
  making them unavailable for assignment.  Note this conversion will occur
  whether or not the server is compiled for failover.
  [ISC-Bugs #36960]

- Fixed a small issue in the treatment of hosts in the inform processing
  that could cause the response to an inform to include information from
  the wrong scope.  The two examples we've heard of are getting subnet
  instead of group information associated with a host entry, or getting
  global information instead of subnet if the host entry was built via
  omapi.  Thanks to Julien Soula at University of Lille for finding the
  bug and supplying a patch.
  [ISC-Bugs #35712]

- Avoid calling pool_timer() recursively from supersede_lease().  This could
  result in leases changing state incorrectly or delaying the running of the
  leae expiration code.
  [ISC-Bugs #38002]

- Move the check for a PID file and process to be before we rewrite the
  lease file.  This avoids the possibility of starting a second instance
  of a server which changes the current lease file confusing the first
  instance.  This check is only included if the admin hasn't disabled PID
  [ISC-Bugs #38078]
  [ISC-Bugs #38143]

- In the client code change the way preferred_life and max_life are printed
  for environment variables to be unsigned rather than signed.
  Thanks to Jiri Popelka at Red Hat for the bug report and patch.
  [ISC-Bugs #37084]

- Modified linux packet handling such that packets received via VLAN are now
  seen only by the VLAN interface. Prior to this, such packets were seen by
  both the VLAN interface and its parent (physical) interface, causing the
  server to respond to both.  Note this remains an issue for non-Linux OSs.
  Thanks to Jiri Popelka at Red Hat for the patch.
  [ISC-Bugs #37415]
  [ISC-Bugs #37133]
  [ISC-Bugs #36668]
  [ISC-Bugs #36652]

- Log content has been changed to more directly suggest that admins should
  check for multiple IPv6 clients attempting to use the same DUID when only
  abandoned addresses are available.  Debug level logging will now emit counts
  of the total number of, in-use, and abandoned addresses in a shared subnet
  when the server finds no addresses available for a given DUID.  Lastly,
  threshold logging is now automatically disabled for shared subnets whose
  total number of possible addresses exceeds (2^64)-1.
  [ISC-Bugs #26376]
  [ISC-Bugs #38131]

- Added a global parameter, prefix-length-mode, which may be used to determine
  how the server uses a non-zero value for prefix-length supplied by clients
  when soliciting DHCPv6 prefixes.  The server supports selection modes of:
  ignore, prefer, exact, minimum and maximum which are described in detail in
  the server man pages.  The prior behavior of the server was to only offer a
  prefix whose length exactly matched the prefix-length value requested. If
  no such prefixes were available, the server returned a status of none
  available.  Note the default mode, "exact", provides this same behavior.
  [ISC-Bugs #36780]
  [ISC-Bugs #32228]

- Corrected inconsistencies in dhcrelay's setting the upper interface hop count
  limit such that it now sets it to 32 when the upstream address is a multicast
  address per RFC 3315 Section 20. Prior to this if the -u argument preceded
  the -l argument on the command line or if the same interface was specified
  for both; the logic to set the hop limit count for the upper interface was
  skipped.  This caused the hop count limit to be set to the default value
  (typically 1) in the outbound upstream packets.
  [ISC-Bugs #37426]

            Changes since 4.3.1b1

- Modify the linux and openwrt dhclient scripts to process information
  from a stateless request.  Thanks to Jiri Popelka at Red Hat for the
  bug report and patch.
  [ISC-Bugs #36102]

- Remove more unused RCSID tags.  These weren't noticed in 4.3 as
  the code isn't used anymore but we remove them here to keep the
  code consistent across versions.
  [ISC-Bugs #36451]

            Changes since 4.3.0

- Tidy up several small tickets.
  Correct parsing of DUID from config file, previously the LL type
  was put in the wrong place in the DUID string.
  [ISC-Bugs #20962]
  Add code to parse "do-forward-updates" as well as "do-forward-update"
  Thanks to Jiri Popelka at Red Hat.
  [ISC-Bugs #31328]
  Remove log_priority as it isn't currently used.
  [ISC-Bugs #33397]
  Increase the size of the buffer used for reading interface information.
  [ISC-Bugs #34858]

- Remove an extra set of the msg_controllen variable.
  [ISC-Bugs #21035]

- Add a more understandable error message if a configuration attempts
  to add multiple keys for a single zone.  Thanks to a patch from Jiri
  Popelka at Red Hat.
  [ISC-Bugs #31892]

- Fix some minor issues in the dst code.
  [ISC-Bugs #34172]

- Properly #ifdef functions so that the code can compile without NSUPDATE.
  [ISC-Bugs #35058]

- Update the partner's stos (start time of state, basically when we last
  heard from this partner) field when updating the state in failover.
  [ISC-Bugs #35549]

- Modify the overload processing to allow space for the remote agent ID.
  [ISC-Bugs #35569]
  Handle the ordering of the SUBNET_MASK option even if it is the last
  option in the list.
  [ISC-Bugs #24580]

- Remove the code that allows a server to follow RFC3315 instead of
  the subsequent errata from August 2010 when determining which IAs
  to include if no addresses will be assigned.
  [ISC-Bugs #28938]

- Remove unused RCSID tags.
  [ISC-Bugs #35846]

- Correct the v6 client timing code.  When doing the timing backoff
  for MRT limit it to MRD.
  Thanks to Jiri Popelka at Red Hat for the bug report and fix.
  [ISC-Bugs #21238

- Add a log entry when killing a client and remove the PID files
  when a server, relay or client are killed.
  [ISC-Bugs #16970]
  [ISC-Bugs #17258]

- Some minor cleanups in the client code.
  In addition to checking for dhcpc check for bootpc in the services list.
  [ISC-Bugs #18933]
  Correct the client code to only try to get a lease once when the
  given the "-1" argument.
  Thanks to Jiri Popelka at Red Hat for the bug report and fix.
  [ISC-Bugs #26735]
  When asked for the version don't send the output to syslog.
  [ISC-Bugs #29772]
  Add the next server information to the environment variables for
  use by the client script.  In order to avoid changing the client
  lease file the next server information isn't written to it.
  Thanks to Tomas Hozza at Red Hat for the suggestion and a prototype fix.
  [ISC-Bugs #33098]

- Several updates to the dhcp server code.
  When not in quiet mode print out the files being used.
  [ISC-Bugs #17551]
  As accessing some pid files may require privileges move the dropping
  of permission bits due to the paranoia patch to be after the pid code.
  Thanks to Jiri Popelka at Red Hat for the bug report and fix.
  [ISC-Bugs #25806]
  When processing a "--version" request don't output the version information
  to syslog.

- Add the "enable-log-pid" build option to the configure script.  When enabled
  this causes the client, server and relay programs to include the PID
  number in syslog messages.
  Thanks to Marius Tomaschewski for the suggestion and proto-patch.
  [ISC-Bugs #29713]

- Add a #define to specify the prefix length used when a client attempts
  to configure an address.  This can be modified by editing includes/site.h.
  By default it is set to 64.  While 128 might be a better choice it would
  also be a change for currently running systems, so we have left it at 64.
  [ISC-Bugs #DHCP-2]

- Add a run time option to the client "-df" to allow the administrator to
  point to a second lease file the client can search for a DUID.  This can
  be used to allow a v4 and a v6 instance of the client to share a DUID.
  The second file will only be searched if there isn't a DUID in the main
  lease file and the DUID will be written out to the main lease file.
  [ISC-Bugs #34886]

- Have the client fsync the lease file to avoid lease corruption if the
  client hibernates or otherwise shuts down.
  [ISC-Bugs #35894]

- Add a check for L2VLAN in bpf.c to help support VLAN interfaces
  Thanks to Steinar Haug for the suggestion.
  [ISC-Bugs #36033]

- Modify the handling of the resolv.conf file to allow the DHCP
  process to start up even if the resolv.conf file has problems.
  [ISC-Bugs #35989]

- Add threshold logging functionality.  Two new options,
  log-threshold-low and log-threshold-high, indicate to the
  server if and when it should log an error message as addresses
  in a pool are used.
  [ISC-Bugs #34487]

- Add code to properly dereference a pointer in the dhclient code
  on an error condition.
  [ISC-Bugs #36194]

- Add code to help clean up soft leases.
  [ISC-Bugs #36304]

- Disable the gentle shutdown functionality until we can determine
  the best way to present it to remove or reduce the side effects.
  [ISC-Bugs #36066]

            Changes since 4.3.0rc1

- None
            Changes since 4.3.0b1

- Tidy up receive packet processing.
  Thanks to Brad Plank of GTA for reporting the issue and suggesting
  a possible patch.
  [ISC-Bugs #34447]

            Changes since 4.3.0a1

- Modify the message displayed when a process hits a fatal error.
  The new message is much shorter and simply points to the README
  and our website for directions on bug submissions.
  [ISC-Bugs #24789]

- Handle an absent resolv.conf file better.
  [ISC-Bugs #35194]

            Changes since 4.2.0 (new features)

- If a client renews before 'dhcp-cache-threshold' percent of its lease
  has elapsed (default 25%), the server will reuse the allocated lease
  (provide a lease within the currently allocated lease-time) rather
  than extend or renew the lease.  This absolves the server of needing
  to perform an fsync() operation on the lease database before reply,
  which improves performance. [ISC-Bugs #22228]
  Updated this patch to support asynchronous DDNS.  If the server is
  attempting to do DDNS on a lease it should be udpated and written to
  disk even if that wouldn't be necessary due to the thresholding.
  [ISC-Bugs #26311]

- The 'no available billing' log line now also logs the name of the last
  matching billing class tried before failing to provide a billing.
  [ISC-Bugs #21759]

- A problem with missing get_hw_addr function when --enable-use-sockets
  was used is now solved on GNU/Linux, BSD and GNU/Hurd systems. Note
  that use-sockets feature was not tested on those systems. Client and
  server code no longer use MAX_PATH constant that is not defined on
  GNU/Hurd systems. [ISC-Bugs #25979]

- Add a perl script in the contrib directory,, which
  can parse v4 lease files and output the lease information in a more
  human friendly manner.  This was written by Christian Hammers with
  some updates by vom and ISC.  This is contributed code and is not
  supported by ISC; however it may be useful to some users.
  [ISC-Bugs #20680]

- Add support in v6 for on-commit, on-expire and on-release.
  [ISC-Bugs #27912]

- Add support for using classes with v6.
  [ISC-Bugs #26510]

- Update the DDNS code to current standards and allow for sharing
  of DDNS entries between v4 and v6 clients.  The new code is used
  if the ddns-update-style is set to "standard", the older code is
  still available if ddns-update-style is set to "interim".  The
  oldest DDNS code "ad-hoc" has been removed.  Thanks to Thomas Pegeot
  who submitted a patch for this issue.  This patch is based on
  that work with some modifications.
  [ISC-Bugs #21139]

- Add a configuration option to the server to suppress using fsync().
  Enabling this option will mean that fsync() is never called.  This
  may provide better performance but there is also a risk that a lease
  will not be properly written to the disk after it has been issued
  to a client and before the server stops.  Using this option is
  not recommended.
  [ISC-Bugs #34810]

- Add some logging statements to indicate when the server is ready
  to serve.  One statement is emitted after the server has finished
  reading its files and is about to enter the dispatch loop.
  This is "Server starting service.".
  The second is emitted when a server determines that both it and
  its failover peer are in the normal state.
  This is "failover peer <name>: Both servers normal."
  [ISC-Bugs #33208]

- Add support for accessing options from v6 relays.  The v6relay
  statement allows the administrator to choose which relay to
  use when searching for an option, see the dhcp-options man page
  for a description.  The host-identifier option has also been
  updated to support the use of relay options, see the dhcpd.conf
  man page for a description.
  [ISC-Bugs #19598]

- When doing DDNS if there isn't an appropriate zone statement attempt
  to find a reasonable nameserver via a DNS resolver.  This restores
  some functionality that was lost in the transition to asynchronous
  DDNS.  Due to the lack of security and increase in fragility of the
  system when using this feature we strongly recommend the use of
  appropriate zone statements rather than using this functionality.
  [ISC-Bugs #30461]

- Add support for specifying the address from which to send
  DDNS updates on the DHCP server.  There are two new options
  "ddns-local-address4" and "ddns-local-address6" that each take
  one instance of their respective address types.
  [ISC-Bugs #34779]

- Add ignore-client-uids option in the server.  This option causes
  the server to not record a client's uid in its lease.  This
  violates the specification but may also be useful when a client
  can dual boot using different client ids but the same mac address.
  Thank you to Brian De Wolf at Cal Poly Pomona for the patch.
  [ISC-Bugs #32427]
  [ISC-Bugs #35066]

- Extend the DHCPINFORM processing to honor the subnet selection option
  and take host declarations into account.
  Thanks to Christof Chen for testing and submitting the patch.
  [ISC-Bugs #35015]

- Extend the hardware expression to look into the lease structure
  for a hardware address if there is no packet.  This allows the
  server to find the hardware address during on-expiry processing.
  [ISC-Bugs #24584]

- Add definitions for some options that have been specified by the IETF.
  [ISC-Bugs #29268]
  [ISC-Bugs #35198]

            Changes since 4.2.0 (bug fixes)

- When using 'ignore client-updates;', the FQDN returned to the client
  is no longer truncated to one octet.

- Cleaned up an unused hardware address variable in nak_lease().

- Manpage entries for the ia-pd and ia-prefix options were updated to
  reflect support for prefix delegation.

- Cleaned up some compiler warnings

- An optimization described in the failover protocol draft is now included,
  which permits a DHCP server operating in communications-interrupted state
  to 'rewind' a lease to the state most recently transmitted to its peer,
  greatly increasing a server's endurance in communications-interrupted.
  This is supported using a new 'rewind state' record on the dhcpd.leases
  entry for each lease.

- Fix the trace code which was broken by the changes to the DDNS code.

- Update the fsync code to work with the changes to the DDNS code.  It now
  uses a timer instead of noticing if there are no more packets to process.

- When constructing the DNS name structure from a text string append
  the root to relative names.  This satisfies a requirement in the DNS
  library that names be absolute instead of relative and prevents DHCP
  from crashing.  [ISC-Bugs #21054]

- "The LDAP Patch" that has been circulating for some time, written by
  Brian Masney and S.Kalyanasundraram and maintained for application to
  the DHCP-4 sources by David Cantrell has been included.  Please be
  advised that these sources were contributed, and do not yet meet the
  high standards we place on production sources we include by default.
  As a result, the LDAP features are only included by using a compile-time
  option which defaults off, and if you enable it you do so under your
  own recognizance.  We will be improving this software over time.
  [ISC-Bugs #17741]

- Prohibit including lease time information in a response to a DHCP INFORM.
  [ISC-Bugs #21092]

! Accept a client id of length 0 while hashing.  Previously the server would
  exit if it attempted to hash a zero length client id, providing attackers
  with a simple denial of service attack.  [ISC-Bugs #21253]
  CERT: VU#541921 - CVE: CVE-2010-2156

- A memory leak in ddns processing was closed.  [ISC-Bugs #21377]

- Modify the exception handling for initial context creation.  Previously
  we would try and clean up before exiting.  This could present problems
  when the cleanup required part of the context that wasn't available.  It
  also didn't do much as we exited afterwards anyway.   Now we simply log
  the error and exit. [ISC-Bugs #21093]

- A bug was fixed that could cause the DHCPv6 server to advertise/assign a
  previously allocated (active) lease to a client that has changed subnets,
  despite being on different shared networks.  Dynamic prefixes specifically
  allocated in shared networks also now are not offered if the client has
  moved.  [ISC-Bugs #21152]

- Add some debugging output for use with the DDNS code. [ISC-Bugs #20916]

- Fix the trace code to handle timing events better and to truncate a file
  before using instead of overwriting it.  [ISC-Bugs #20969]

- Modify the determination of the default TTL to use for DDNS updates.
  The user may still configure the ttl via ddns-ttl.  The default for
  both v4 and v6 is now 1/2 the (preferred) lease time with a limit.  The
  previous defaults (1/2 lease time without a limit for v4 and a default
  value for v6) may be used by defining USE_OLD_DDNS_TTL in site.h
  [ISC-Bugs #21126]

- libisc/libdns is now brought up to version 9.7.1rc1.  This corrects
  three reported flaws in ISC DHCP;

  o DHCP processes (dhcpd, dhclient) fail to start if one of either the
    IPv4 or IPv6 address families is not present.  [ISC-Bugs #21122]

  o Assertion failure when attempting to cancel a previously running DDNS
    update.  [ISC-Bugs #21133]

  o Compilation failure of libisc/libdns due to the use of a flexible
    array member.  [ISC-Bugs #21316]

- Add declaration for variable in debug code in alloc.c.  [ISC-Bugs #21472]

- Documentation cleanup covering multiple tickets
  [ISC-Bugs #20265] [ISC-Bugs #20259] minor cleanup
  [ISC-Bugs #20263] add text describing some default values
  [ISC-Bugs #20193] single quotes at the start of a line indicate a control
  line to nroff, escape them if we actually want a quote.
  [ISC-Bugs #18916] sync the pointer to web pages amongst the different docs

- 'get-host-names true;' now also works even if 'use-host-decl-names true;'
  was also configured.  The nature of this repair also fixes another
  error; the host-name supplied by a client is no longer overridden by a
  reverse lookup of the lease address.  Thanks to a patch from Wilco Baan
  Hofman supplied to us by the Debian package maintenance team.
  [ISC-Bugs #21691] {Debian Bug#509445}

- The .TH tag for the dhcp-options manpage was typo repaired
  thanks to a report from jidanni and the Debian package maintenance
  team.  [ISC-Bugs #21676] {Debian Bug#563613}

- More documentation changes - primarily to put the options in the dhclient
  and dhcpd man pages into the standard form.  Thanks in part to a patch
  from David Cantrell at Red Hat.
  [ISC-Bugs #20264] and parts of [ISC-Bugs #17744] dhclient.8 changes

- Add code to clear the pointer to an object in an OMAPI handle when the
  object is freed due to a dereference.  [ISC-Bugs #21306]

- Fixed a bug that leaks host record references onto lease structures,
  causing the server to apply configuration intended for one host to any
  other innocent clients that come along later.  [ISC-Bugs #22018]

- Minor code fixes
  [ISC-Bugs #19566] When trying to find the zone for a name for ddns allow
  the name to be at the apex of the zone.
  [ISC-Bugs #19617] Restrict length of interface name read from command line
  in dhcpd - based on a patch from David Cantrell at Red Hat.
  [ISC-Bugs #20039] Correct some error messages in dhcpd.c
  [ISC-Bugs #20070] Better range check on values when creating a DHCID.
  [ISC-Bugs #20198] Avoid writing past the end of the field when adding
  overly long file or server names to a packet and add a log message
  if the configuration supplied overly long names for these fields.
  Thanks to Martin Pala.
  [ISC-Bugs #21497] Add a little more randomness to rng seed in client
  thanks to a patch from Jeremiah Jinno.

- Correct error handling in DLPI [ISC-Bugs #20378]

- Remove __sun__ and __hpux__ typedefs in osdep.h as they are now being
  checked in configure.  [ISC-Bugs #20443]

- Modify how the cmsg header is allocated the v6 send and received routines
  to compile on more compilers.  [ISC-Bugs #20524]

- When parsing a domain name free the memory for the name after we are
  done with it.  [ISC-Bugs #20824]

- Add an elapsed time option to the release message and refactor the
  code to move most of the common code to a single routine.
  [ISC-Bugs #21171].

- Two identical log messages for commit_leases() have been disambiguated.
  [ISC-Bugs #18915]

- Parse date strings more properly - the code now handles semi-colons in
  date strings correctly.  Thanks to a patch from Jiri Popelka at Red Hat.
  [ISC-Bugs #21501, #20598]

- Fixes to lease input and output.
  [ISC-Bugs #20418] - Some systems don't support the "%s" argument to
  strftime, paste together the same string using mktime instead.
  [ISC-Bugs #19596] - When parsing iaid values accept printable
  [ISC-Bugs #21585] - Always print time values in omshell as hex
  instead of ascii if the values happen to be printable characters.

- Minor changes for scripts, and Makefiles
  [ISC-Bugs #19147] Use domain-search instead of domain-name in manual and
                    example conf file.  Thanks to a patch from David Cantrell
                    at Red Hat.
  [ISC-Bugs #19761] Restore address when doing a rebind in DHCPv6
  [ISC-Bugs #19945] Properly close the quote on some arguments.
  [ISC-Bugs #20952] Add 64 bit types to
  [ISC-Bugs #21308] Add "PATH=" to CLIENT_PATH envrionment variable

- Update the code to parse dhcpv6 lease files to accept a semi-colon at
  the end of the max-life and preferred-life clauses.  In order to be
  backwards compatible with older lease files not finding a semi-colon
  is also accepted.  [ISC-Bugs #22303].

! Handle a relay forward message with an unspecified address in the
  link address field.  Previously such a message would cause the
  server to crash.  Thanks to a report from John Gibbons.  [ISC-Bugs #21992]
  CERT: VU#102047 CVE: CVE-2010-3611

- ./configure on longer searches for -lcrypto to explicitly link against.
  This fixes a bug where 'dhclient' would have shared library dependencies
  on '/usr/lib'.  [ISC-Bugs #21967]

- Handle pipe failures more gracefully.  Some OSes pass a SIGPIPE
  signal to a process and will kill the process if the signal isn't
  caught.  This patch adds code to turn off the SIGPIPE signal via
  a setsockopt() call.  The signal is already being ignored as part
  of the ISC library.  [ISC-Bugs #22269]

- Restore printing of values in omshell to the style pre 21585.  For
  21585 we changed the print routines to always display time values
  as a hex list.  This had a side effect of printing all data strings
  as a hex list.  We shall investigate other ways of displaying time
  values more usefully.  [ISC-Bugs #22626]

! Fix the handling of connection requests on the failover port.
  Previously a connection request from a source that wasn't
  listed as a failover peer would cause the server to become
  non-responsive.  Thanks to a report from Brad Bendily,
  [ISC-Bugs #22679]
  CERT: VU#159528 CVE: CVE-2010-3616

- Don't pass the ISC_R_INPROGRESS status to the omapi signal handlers.
  Passing it through to the handlers caused the omshell program to fail
  to connect to the server.  [ISC-Bugs #21839]

- Fix the paranthesis in the code to process configuration statements
  beginning with "auth".  The previous arrangement caused
  "auto-partner-down" to be processed incorrectly.  [ISC-Bugs #21854]

- Limit the timeout period allowed in the dispatch code to 2^^32-1 seconds.
  Thanks to a report from Jiri Popelka at Red Hat.
  [ISC-Bugs #22033], [Red Hat Bug #628258]

- When processing the format flags for a given option consume the
  flag indicating an optional value correctly.  A symptom of this
  bug was an infinite loop when trying to parse the slp-service-scope
  option.  Thanks to a patch from Marius Tomaschewski.
  [ISC-Bugs #22055]

- Disable the use of kqueue in the ISC library.  This avoids a problem
  between the fork and socket code that caused the dhcpd process to
  use all available cpu if the program daemonized itself.
  [ISC-Bugs #21911]

! When processing a request in the DHCPv6 server code that specifies
  an address that is tagged as abandoned (meaning we received a
  decline request for it previously) don't attempt to move it from
  the inactive to active pool as doing so can result in the server
  crashing on an assert failure.  Also retag the lease as active
  and reset its timeout value.
  [ISC-Bugs #21921]

- Removed the restriction on using IPv6 addresses in IPv4 mode.  This
  allows IPv4 options which contain IPv6 addresses to be specified.  For
  example the 6rd option can be specified and used like this:
  [ISC-Bugs #23039]

    option 6rd code 212 = { integer 8, integer 8,
                ip6-address, array of ip-address };
    option 6rd 16 10 2001::,;

- Handle some DDNS corner cases better.  Maintain the DDNS transaction
  information when updating a lease and cancel any existing transactions
  when removing the ddns information.
  [ISC-Bugs #23103]

- Some fixes for LDAP
  [ISC-Bugs #21783] - Include lber library when building ldap
  [ISC-Bugs #22888] - Enable the ldap code when buidling common
  The above fixes are from Jiri Popelka at Red Hat.

- Modify the dlpi code to accept getmsg() returning a positive value.
  [ISC-Bugs #22824]

! In dhclient check the data for some string options for
  reasonableness before passing it along to the script that
  interfaces with the OS.
  [ISC-Bugs #23722]
  CVE: CVE-2011-0997

- DHCPv6 server now responds properly if client asks for a prefix that
  is already assigned to a different client. [ISC-Bugs #23948]

- Add the option "--no-pid" to the client, relay and server code,
  to disable writing a pid file.  Add the option "-pf pidfile"
  to the relay to allow the user to supply the pidfile name at
  runtime.  Add the "with-relay6-pid-file" option to configure
  to allow the user to supply the pidfile name for the relay
  in v6 mode at configure time.
  [ISC-Bugs #23351] [ISC-Bugs #17541]

- 'dhclient' no longer waits a random interval after first starting up to
  begin in the INIT state.  This conforms to RFC 2131, but elects not to
  implement a 'SHOULD' direction in section 4.1. The goal of this change
  is to start up faster. [ISC-Bugs #19660]

- Added 'initial-delay' parameter that specifies maximum amount of time
  before client goes to the INIT state. The default value is 0. In previous
  versions of the code client could wait up to 5 seconds. The old behavior
  may be restored by using 'initial-delay 5;' in the client config file.
  [ISC-Bugs #19660]

- ICMP ping-check should now sit closer to precisely the number of seconds
  configured (or default 1), due to making use of the new microsecond
  scale timer internally to dhcpd.  This corrects a bug where the server
  may immediately timeout an ICMP ping-check if it was made late in the
  current second. [ISC-Bugs #19660]

- The DHCP client will schedule renewal and rebinding events in
  microseconds if the DHCP server provided a lease-time that would result
  in sub-1-second timers.  This corrects a bug where a 2-second or lower
  lease-time would cause the DHCP client to enter an infinite loop by
  scheduling renewal at zero seconds. [ISC-Bugs #19660]

- Client lease records are recorded at most once every 15 seconds.  This
  keeps the client from filling the lease database disk quickly on very small
  lease times. [ISC-Bugs #19660]

- To defend against RFC 2131 non-compliant DHCP servers which fail to
  advertise a lease-time (either mangled, or zero in value) the DHCP
  client now adds the server to the reject list ACL and returns to INIT
  state to hopefully find an RFC 2131 compliant server (or retry in INIT
  forever). [ISC-Bugs #19660]

- Parameters configured to evaluate from user defined function calls can
  now be correctly written to dhcpd.leases (as on 'on events' or dynamic
  host records inserted via OMAPI).  [ISC-Bugs #22266]

- If a 'next-server' parameter is configured in a dynamic host record via
  OMAPI as a domain name, the syntax written to disk is now correctly parsed
  upon restart.  [ISC-Bugs #22266]

- The DHCP server now responds to DHCPLEASEQUERY messages from agents using
  IP addresses not covered by a subnet in configuration.  Whether or not to
  respond to such an agent is still governed by the 'allow leasequery;'
  configuration parameter, in the case of an agent not covered by a configured
  subnet the root configuration area is examined. Server now also returns
  vendor-class-id option, if client sent it. [ISC-Bugs #21094]

- Documentation fixes
  [ISC-Bugs #17959] add text to AIX section describing how to have it send
  responses to the all-ones address.
  [ISC-Bugs #19615] update the includes in dhcpctl/dhcpctl.3 to be more correct
  [ISC-Bugs #20676] update dhcpd.conf.5 to include the RFC numbers for DDNS

- Relay no longer crashes, when DHCP packet is received over interface without
  any IPv4 address assigned. Also extended logging message about discarding
  packets with invalid hlen with information about relevant interface name.
  [ISC-Bugs #22409]

- Relay now properly logs that packet was received over interface without
  global IPv6 address [ISC-Bugs #24070]

- Linux Packet Filter interface improvement. sockaddr_pkt structure is used,
  rather than sockaddr. Packet ethertype is now forced to ETH_P_IP.
  [ISC-Bugs #18975]

- Minor code cleanups - but note port change for #23196
  [ISC-Bugs #23470] - Modify when an ignore return macro is defined to
  handle unsed error return warnings for more versions of gcc.
  [ISC-Bugs #23196] - Modify the reply handling in the server code to
  send to a specified port rather than to the source port for the incoming
  message.  Sending to the source port was test code that should have
  been removed.  The previous functionality may be restored by defining
  REPLY_TO_SOURCE_PORT in the includes/site.h file.  We suggest you don't
  enable this except for testing purposes.
  [ISC-Bugs #22695] - Close a file descriptor in an error path.
  [ISC-Bugs #19368] - Tidy up variable types in validate_port.

- Code cleanup: remove obsolete PROTO, KandR, INLINE and ANSI_DECL macros
  [ISC-Bugs #13151]

- Compilation problem with gcc4.5 and omshell.c resolved. [ISC-Bugs #23831]

- Client Script fixes
  [ISC-Bugs #23045] Typos in client/scripts/openbsd
  [ISC-Bugs #23565] In the client scripts add a zone id (interface id) if
  the domain search address is link local.
  [ISC-Bugs #1277] In some of the client scripts add code to handle the
  case of the default router information being changed without the address
  being changed.

- Documentation cleanup
  [ISC-Bugs #23326] Updated References document, several man page updates

- Server no longer complains about NULL pointer when configured
  server-identifier expression fails to evaluate. [ISC-Bugs #24547]

- Convert ISC_R_INPROGRESS status to ISC_R_SUCCESS when called from other
  than the dispatch handler.  This fixes an issue where omshell, when
  run from the same platform as the server, would appear to fail to
  connect.  This is a companion to #21839.  [ISC-Bugs #23592]

- Enlarge the buffer size used by the Omshell code and some of the
  print routines to allow for greater than 60 characters or, when
  printing as hex strings, 20 characters.  [ISC-Bugs #22743]

- In Solaris 11 switch to using sockets instead of DLPI, thanks
  to a patch form Oracle.  [ISC-Bugs #24634].

- Strict checks for content of domain-name DHCPv4 option can now be
  configured during compilation time. Even though RFC2132 does not allow
  to store more than one domain in domain-name option, such behavior is
  now enabled by default, but this may change some time in the future.
  See ACCEPT_LIST_IN_DOMAIN_NAME define in includes/site.h.
  [ISC-Bugs #24167]

- DNS Update fix. A misconfigured server could crash during DNS update
  processing if the configuration included overlapping pools or
  multiple fixed-address entries for a single address.  This issue
  affected both IPv4 and IPv6. The fix allows a server to detect such
  conditions, provides the user with extra information and recommended
  steps to fix the problem.  If the user enables the appropriate option
  in site.h then server will be terminated
  [ISC-Bugs #23595]

! Two packets were found that cause a server to halt.  The code
  has been updated to properly process or reject the packets as
  appropriate.  Thanks to David Zych at University of Illinois
  for reporting this issue.  [ISC-Bugs #24960]
  One CVE number for each class of packet.

- Fix the code that checks for an existing DDNS transaction to cancel
  when removing DDNS information, so that we will continue with the
  processing if we have a lease even if it doesn't have an outstanding
  transaction.  [ISC-Bugs #24682]

- Add AM_MAINTAINER_MODE to to avoid rebuilding
  configuration files.  [ISC-Bugs #24107]

- Add support for passing DDNS information to a DNS server over
  an IPv6 address.  [ISC-Bugs #22647]

- Enhanced patch for 23595 to handle IPv4 fixed addresses more
  cleanly.  [ISC-Bugs #23595]

! Add a check for a null pointer before calling the regexec function.
  Without this check we could, under some circumstances, pass
  a null pointer to the regexec function causing it to segfault.
  Thanks to a report from BlueCat Networks.
  [ISC-Bugs #26704].
  CVE: CVE-2011-4539

! Modify the DDNS handling code.  In a previous patch we added logging
  code to the DDNS handling.  This code included a bug that caused it
  to attempt to dereference a NULL pointer and eventually segfault.
  While reviewing the code as we addressed this problem, we determined
  that some of the updates to the lease structures would not work as
  planned since the structures being updated were in the process of
  being freed: these updates were removed.  In addition we removed an
  incorrect call to the DDNS removal function that could cause a failure
  during the removal of DDNS information from the DNS server.
  Thanks to Jasper Jongmans for reporting this issue.
  [ISC-Bugs #27078]
  CVE: CVE-2011-4868

- Fixed the code that checks if an address the server is planning
  to hand out is in a reserved range.  This would appear as
  the server being out of addresses in pools with particular ranges.
  [ISC-Bugs #26498]

- In the DDNS code handle error conditions more gracefully and add more
  logging code.  The major change is to handle unexpected cancel events
  from the DNS client code.
  [ISC-Bugs #26287]

- Tidy up the receive calls and eliminate the need for found_pkt.
  [ISC-Bugs #25066]

- Add support for Infiniband over sockets to the server and
  relay code.  We've tested this on Solaris and hope to expand
  support for Infiniband in the future.  This patch also corrects
  some issues we found in the socket code.
  [ISC-Bugs #24245]

- Add a compile time check for the presence of the noreturn attribute
  and use it for log_fatal if it's available.  This will help code
  checking programs to eliminate false positives.
  [ISC-Bugs #27539]

- Fixed many compilation problems ("set, but not used" warnings) for
  gcc 4.6 that may affect Ubuntu 11.10 users. [ISC-Bugs #27588]

- Modify the code that determines if an outstanding DDNS request
  should be cancelled.  This patch results in cancelling the
  outstanding request less often.  It fixes the problem caused
  by a client doing a release where  the TXT and PTR records
  weren't removed from the DNS.
  [ISC-BUGS #27858]

- Use offsetof() instead of sizeof() to get the sizes for dhcpv6_relay_packet
  and dhcpv6_packet in several more places.  Thanks to a report from
  Bruno Verstuyft and Vincent Demaertelaere of Excentis.
  [ISC-Bugs #27941]

- Remove outdated note in the description of the bootp keyword about the
  option not satisfying the requirement of failover peers for denying
  dynamic bootp clients.
  [ISC-bugs #28574]

- Multiple items to clean up IPv6 address processing.
  When processing an IA that we've seen check to see if the
  addresses are usable (not in use by somebody else) before
  handing it out.
  When reading in leases from the file discard expired addresses.
  When picking an address for a client include the IA ID in
  addition to the client ID to generally pick different addresses
  for different IAs.
  [ISC-Bugs #23138] [ISC-Bugs #27945] [ISC-Bugs #25586]
  [ISC-Bugs #27684]

- Remove unnecessary checks in the lease query code and clean up
  several compiler issues (some dereferences of NULL and treating
  an int as a boolean).
  [ISC-Bugs #26203]

- Fix the NA and PD allocation code to handle the case where a client
  provides a preference and the server doesn't have any addresses or
  prefixes available.  Previously the server ignored the request with
  this patch it replies with a NoAddrsAvail or NoPrefixAvail response.
  By default the code performs according to the errata of August 2010
  for RFC 3315 section 17.2.2; to enable the previous style see the
  section on RFC3315_PRE_ERRATA_2010_08 in includes/site.h.  This option
  may be removed in the future.
  Thanks to Jiri Popelka at Red Hat for the patch.
  [ISC-Bugs #22676]

- Fix up some issues found by static analysis.
  A potential memory leak and NULL dereference in omapi.
  The use of a boolean test instead of a bitwise test in dst.
  [ISC-Bugs #28941]

- Rotate the lease file when running in v6 mode.
  Thanks to Christoph Moench-Tegeder at Astaro for the
  report and the first version of the patch.
  [ISC-Bugs #24887]

- Correct code to calculate timing values in client to compare
  rebind value to infinity instead of renew value.
  Thanks to Chenda Huang from H3C Technologies Co., Limited
  for reporting this issue.
  [ISC-Bugs #29062]

- Fix some issues in the code for parsing and printing options.
  [ISC-Bugs #22625] - properly print options that have several fields
  followed by an array of something for example "fIa"
  [ISC-Bugs #27289] - properly parse options in declarations that have
  several fields followed by an array of something for example "fIa"
  [ISC-Bugs #27296] - properly determine if we parsed a 16 or 32 bit
  value in evaluate_numeric_expression (extract-int).
  [ISC-Bugs #27314] - properly parse a zero length option from
  a lease file.  Thanks to Marius Tomaschewski from SUSE for the report
  and prototype patch for this ticket as well as ticket 27289.

! Previously the server code was relaxed to allow packets with zero
  length client ids to be processed.  Under some situations use of
  zero length client ids can cause the server to go into an infinite
  loop.  As such ids are not valid according to RFC 2132 section 9.14
  the server no longer accepts them.  Client ids with a length of 1
  are also invalid but the server still accepts them in order to
  minimize disruption.  The restriction will likely be tightened in
  the future to disallow ids with a length of 1.
  Thanks to Markus Hietava of Codenomicon CROSS project for the
  finding this issue and CERT-FI for vulnerability coordination.
  [ISC-Bugs #29851]
  CVE: CVE-2012-3571

! When attempting to convert a DUID from a client id option
  into a hardware address handle unexpected client ids properly.
  Thanks to Markus Hietava of Codenomicon CROSS project for the
  finding this issue and CERT-FI for vulnerability coordination.
  [ISC-Bugs #29852]
  CVE: CVE-2012-3570

! A pair of memory leaks were found and fixed.  Thanks to
  Glen Eustace of Massey University, New Zealand for finding
  this issue.
  [ISC-Bugs #30024]
  CVE: CVE-2012-3954

- Existing legacy unit-tests have been migrated to Automated Test
  Framework (ATF). Several new tests have been developed. To enable
  unit-tests, please use --with-atf in configure script. A Developer's
  Guide has been added. To generate it, please use make devel in
  the doc directory. It is currently in early stages of development,
  but is expected to grow in the near future. [ISC-Bugs 25901]

! An issue with the use of lease times was found and fixed.  Making
  certain changes to the end time of an IPv6 lease could cause the
  server to abort.  Thanks to Glen Eustace of Massey University,
  New Zealand for finding this issue.
  [ISC-Bugs #30281]
  CVE: CVE-2012-3955

- Update the memory leakage debug code to work with v6.
  [ISC-Bugs #30297]

- Relax the requirements for deleting an A or AAAA record.
  Previously the DDNS removal code required both the A or AAAA
  record and the TXT record to exist.  This requirement could
  cause problems if something interrupted the removal leaving
  the TXT record alone.  This relaxation was codified in RFC 4703.
  [ISC-Bugs #30734]

- Modify the failover code to handle incorrect peer names
  better.  Previously the structure holding the name might
  have been freed inappropriately in some cases and not
  freed in other cases.
  [ISC-Bugs #30320]

- Add a configure option, enable-secs-byteorder, to deal with
  clients that do the byte ordering on the secs field incorrectly.
  This field should be in network byte order but some clients
  get it wrong.  When this option is enabled the server will examine
  the secs field and if it looks wrong (high byte non zero and low
  byte zero) swap the bytes.  The default is disabled.  This option
  is only useful when doing load balancing within failover.
  [ISC-Bugs #26108]

- Fix a set of issues that were discovered via a code inspection
  tool.  Thanks to Jiri Popelka and Tomas Hozza Red Hat for the logs
  and patches.
  [ISC-Bugs #23833]

- Parsing unquoted base64 strings improved. Parser now properly handles
  strings that contain reserved names. [ISC-Bugs #23048]

- Modify the nak_lease function to make some attempts to find a
  server-identifier option to use for the NAK.
  [ISC-Bugs #25689]

- The client now passes information about the options it requested
  from the server to the script code via environment variables.
  These variables are of the form requested_<option_name>=1 with
  the option name being the same as used in the new_* and old_*
  [ISC-Bugs #29068]

- Add support for a simple check that the server id in a request message
  to a failover peer matches the server id of the server.  This support
  is enabled by editing the file includes/site.h and uncommenting the
  definition for SERVER_ID_CHECK.  The option has several restrictions
  and issues - please read the comment in the site.h file before
  enabling it.
  [ISC-Bugs #31463]

- Tidy up some compiler issues in the debug code.
  [ISC-Bugs #26460]

- Move the dhcpd.conf exmample file to dhcpd.conf.example to avoid
  overwriting the dhcpd.conf file when installing a new version of
  ISC DHCP.  The user will now need to manual copy and edit the
  dhcpd.conf file as desired.
  [ISC-Bugs #19337]

- Check the status value when trying to read from a connection to
  see if it may have been closed.  If it appears closed don't try
  to read from it again.  This avoids a potential busy-wait like
  loop when the peer names are mismatched.
  [ISC-Bugs #31231]

- Remove an unused variable to keep compilers happy.
  [ISC-Bugs #31983]

- Modify test makefiles to be more similar to standard makefiles
  and comment out a currently unused test.
  [ISC-Bugs #32089]

- Address static analysis warnings.
  [ISC-Bugs #33510] [ISC-Bugs #33511]

- Silence benign static analysis warnings.
  [ISC-Bugs #33428]

- Add check for 64-bit package for atf.
  [ISC-Bugs #32206]

- Use newer auto* tool packages and turn on RFC_3542 support on Mac OS.
  [ISC-Bugs #26303]

- Remove a variable when it isn't being used due to #ifdefs to avoid
  a compiler warning on Solaris using GCC.
  [ISC-Bugs #33032]

- Add a check for too much whitespace in a config or lease file.
  Thanks to Paolo Pellegrino for finding the issue and a suggestion
  for the patch.
  [ISC-Bugs #33351]

- Fix several problems with using OMAPI to manipulate class and subclass
  [ISC-Bugs #27452]

- Added a sleep call after killing the old client to allow time
  for the sockets to be cleaned.  This should allow the -r option
  to work more consistently.
  [ISC-Bugs #18175]

- Missing files for ISC DHCP Developer's Guide are now included in
  the release tarballs. To generate this documentation, please use
  make devel command in doc directory. [ISC-Bugs #32767]

- Update client script for use with openwrt.
  [ISC-Bugs #29843]

- Fix the socket handling for DHCPv6 clients to allow multiple instances
  of a client on a single machine to work properly.  Previously only
  one client would receive the packets.  Thanks to Jiri Popelka at Red Hat
  for the bug report and a potential patch.
  [ISC-Bugs #34784]

- Added support for gentle shutdown after signal is received.
  [ISC-Bugs #32692] [ISC-Bugs 34945]

- Enhance the DHCPv6 server logging to include the addresses that are assigned
  to the clients.
  [ISC-Bugs #26377]

- Fix an operation in the DDNS code to be a bitwise instead of logical or.
  [ISC-Bugs #35138]

            Changes since 4.1.0 (new features)

- Failover port configuration can now be left to defaults (port 647) as
  described in the -12 revision of the Failover draft (and assigned by
  IANA).  Thanks in part to a patch from David Cantrell at Red Hat.

- If configured, dhclient may now transmit to an anycast MAC address,
  rather than using a broadcast address.  Thanks to a patch from David
  Cantrell at Red Hat.

- Added client support for setting interface MTU and metric, thanks to
  Roy "UberLord" Marples <>.

- Added client -D option to specify DUID type to send.

- A new failover configuration parameter has been introduced for those
  environments where DHCP servers can be reasonably guaranteed to be
  "down" when the failover TCP socket is severed, "auto-partner-down".
  This parameter is not generally safe, and by default is disabled, so
  please carefully review the documentation of this parameter in the
  dhcpd.conf(5) manpage before determining to use it yourself.

- Added a configuration function, 'gethostname()', which calls the system
  function of the same name and presents the results as a data expression.
  This function can be used to incorporate the system level hostname of
  the system the DHCP software is operating on in responses or queries (such
  as including a failover partner's hostname in a dhcp message or binding
  scope, or having a DHCP client send any system hostname in the host-name or
  FQDN options by default).

- The dhcp-renewal-time and dhcp-rebinding-time options may now be configured
  for DHCPv4 operation and used independently of the dhcp-lease-time
  calculations.  Invalid renew and rebinding times (e.g., greater than the
  determined lease time) are omitted.

- Processing the DHCP to DNS server transactions in an asyncrhonous fashion,
  the DHCP server or client can now continue with its processing while
  awaiting replies from the DNS server.

- The 'hardware [ethernet|etc] ...;' parameter in host records has been
  extended to attempt to match DHCPv6 clients by the last octets of a
  DUID-LL or DUID-LLT provided by the client.

            Changes since 4.1.0 (bug fixes)

- Remove infinite loop in token_print_indent_concat().

- Validate the argument to the -p option.

- The notorious 'option <unknown> ... larger than buffer' log line,
  which is seen in some malformed DHCP client packets, was modified.
  It now logs the universe name, and does not log the length values
  (which are bogus corruption read from the packet anyway).  It also
  carries a hopefully more useful explanation.

- Suppress spurious warnings from configure about --datarootdir

- A bug was fixed that caused the server not to answer some valid Solicit
  and Request packets, if the dynamic range covering any requested addresses
  had been deleted from configuration.

- Update the code to deal with GCC 4.3.  This included two sets of changes.
  The first is to the configuration files to include the use of
  AC_USE_SYSTEM_EXTENSIONS.  The second is to deal with return values that
  were being ignored.

- The db-time-format option was documented in manpages.

- Using reserved leases no longer results in 'lease with binding state
  free not on its queue' error messages, thanks to a patch from Frode

- Fix a build error in dhcrelay, using older versions of gcc with
  dhcpv6 disabled.

- Two uninitialized stack structures are now memset to zero, thanks to a
  patch from David Cantrell at Red Hat.

- Fixed a cosmetic bug where pretty-printing valid domain-search options would
  result in an erroneous error log message ('garbage in format string').

- A bug in DLPI packet transmission (Solaris, HP/UX) that caused the server
  to stop receiving packets is fixed.  The same fix also means that the MAC
  address will no longer appear 'bogus' on DLPI-based systems.

- A bug in select handling was discovered where the results of one select()
  call were discarded, causing the server to process the next select() call
  and use more system calls than required.  This has been repaired - the
  sockets will be handled after the first return from select(), resulting in
  fewer system calls.

- The update-conflict-detection feature would leave an FQDN updated without
  a DHCID (still currently implemented as a TXT RR).  This would cause later
  expiration or release events to fail to remove the domain name.  The feature
  now also inserts the client's up to date DHCID record, so records may safely
  be removed at expiration or release time.  Thanks to a patch submitted by
  Christof Chen.

- Memory leak in the load_balance_mine() function is fixed.  This would
  leak ~20-30 octets per DHCPDISCOVER packet while failover was in use
  and in normal state.

- Various compilation fixes have been included for the memory related
  DEBUG #defines in includes/site.h.

- Fixed Linux client script 'unary operator expected' errors with DHCPv6.

- Fixed setting hostname in Linux hosts that require hostname argument
  to be double-quoted.  Also allow server-provided hostname to
  override hostnames 'localhost' and '(none)'.

- Fixed failover reconnection retry code to continue to retry to reconnect
  rather than restarting the listener.

- Compilation on Solaris with USE_SOCKETS defined in includes/site.h has
  been repaired.  Other USE_ overrides should work better.

- A check for the local flavor of IFNAMSIZ had a broken 'else' condition,
  that probably still resulted in the correct behaviour (but wouldn't use
  a larger defined value provided by the host OS).

- Fixed a bug where an OMAPI socket disconnection message would not result
  in scheduling a failover reconnection, if the link had not negotiated a
  failover connect yet (e.g.: connection refused, asynch socket connect()

- A bug was fixed that caused the 'conflict-done' state to fail to be parsed
  in failover state records.

! A stack overflow vulnerability was fixed in dhclient that could allow
  remote attackers to execute arbitrary commands as root on the system,
  or simply terminate the client, by providing an over-long subnet-mask
  option.  CERT VU#410676 - CVE-2009-0692

- Fixed a bug where relay agent options would never be returned when
  processing a DHCPINFORM.

- Versions 3.0.x syntax with multiple name->code option definitions is now
  supported.  Note that, similarly to 3.0.x, for by-code lookups only the
  last option definition is used.

- Fixed a bug where a time difference of greater than 60 seconds between a
  failover pair could cause the primary to crash on contact with the
  secondary.  Thanks to a patch from Steinar Haug.

- Don't look for IPv6 interfaces on Linux when running in DHCPv4 mode.
  Thanks to patches from Matthew Newton and David Cantrell.

- Secondary servers in a failover pair will now perform ddns removals if
  they had performed ddns updates on a lease that is expiring, or was
  released through the primary.  As part of the same fix, stale binding scopes
  will now be removed if a change in identity of a lease's active client is
  detected, rather than simply if a lease is noticed to have expired (which it
  may have expired without a failover server noticing in some situations).

- A patch supplied by David Cantrell at RedHat was applied that detects
  invalid calling parameters given to the ns_name_ntop() function.
  Specifically, it detects if the caller passed a pointer and size pair
  that causes the pointer to integer-wrap past zero.

! Fixed a fenceposting bug when a client had two host records configured,
  one using 'uid' and the other using 'hardware ethernet'.  CVE-2009-1892

- Fixed the check in the dhcp_interface_signal_handler routine to verify
  the existence of the linked signal handler before calling it.

- Both host and subnet6 configuration groups are now included whether a
  fixed-address6 (DHCPv6) is in use or not.  Host scoped configuration takes
  precedence.  This fixes two bugs, one where host scoped configuration
  would not be included from a non-fixed-address6 host record, and the equal
  and opposite bug where subnet6 scoped configuration would not be used when
  over-riding values were not present in a matching fixed-address6 host

- ./configure now checks to ensure the intX_t and u_intX_t types are defined,
  correcting a compilation failure when using Sun's compiler.

- Modified the handling of a connection to avoid releasing the omapi io
  object for the connection while it is still in use.  One symptom from
  this error was a segfault when a failover secondary attempted to connect
  to the failover primary if their clocks were not synchronized.

- Clean up to allow compilation with gcc 2.95.4 on FreeBSD.  Remove an
  extra semi-colon from common/dns.c and moved setting a variable to NULL
  in server/dhcpv6.c to allow the compiler to decide that the variable
  was always properly set.

            Changes since 4.1.0b1

- A missing "else" in dhcrelay.c could have caused an interface not to
  be recognized.

            Changes since 4.1.0a2

- A cosmetic bug in DHCPDECLINE processing was fixed which caused all
  successful DHCPDECLINEs to be logged as "not found" rather than

- Added configuration file examples for DHCPv6.

- Some failover debugging #defines have been better defined and some
  high frequency messages moved to a deeper debugging symbol.

- The CLTT parameter in failover is now only updated by client activity,
  and not by failover binding updates (taking on the peer's CLTT).

- Failover BNDUPD messages are now discarded if they conflict with an
  update that has been transmitted, but not acknowledged.

- A bug cleaning up unknown-xxx temporary option definitions was fixed.

- Delayed-ack is now a compile-time option, compiled out by default.
  This feature is simply too experimental for right now, and causes
  some problems to some failover installations.  We will revisit this
  in future releases.

- The !inet_pton() call in res_mkupdrec was adjusted to '<= 0' as
  inet_pton returns either 1, 0, or -1.

- A dhclient-script for MacOS X has been included, which enables
  'dhclient -6' support.

- DDNS removal routines were updated so that the DHCID is not removed until
  the client has been deprived of all A and AAAA records (not only the last
  one of either of those).  This resolves a bug where dual stack clients
  would not be able to regain their names after either expiration event.

            Changes since 4.1.0a1

- Corrected list of failover state values in dhcpd man page.

- Fixed a bug that caused some request types to be logged incorrectly.

- Clients that sent a parameter request list containing the
  routers option before the subnet mask option were receiving
  only the latter.  Fixed.

- The server wasn't always sending the FQDN option when it should.

- A partner-down failover server no longer emits 'peer holds all free leases'
  if it is able to newly-allocate one of the peer's leases.

- Fixed a coredump when adding a class via OMAPI.

- Check whether files are zero length before trying to parse them.

- Ari Edelkind's PARANOIA patch has been included and may be compiled in
  via two ./configure parameters, --enable-paranoia and

- ./configure was extended to cover many optional build features, such
  as failover, server tracing, debugging, and the execute() command.

- There is now a default 1/4 of a second scheduled delay between delayed
  fsync()'s, it can be configured by the max-ack-delay configuration

- A bug was fixed where the length of a hostname was miscalculated, so that
  hosts were given odd-looking domain names ("").

- Shared network selection should be done from the innermost relay
  valid link-address field, rather than the outermost.

- Prefix pools are attached to shared network scopes.

- Merged IA_XX related structures.

- Add DHCPv6 files in configure.

- A memory leak when using omapi has been fixed.

- DHCPv6 vendor-class options (VSIO) are now only sent when they appear
  on the DHCPv6 ORO.  This resolves a bug where VSIO options were placed
  in IA_NA encapsulated options fields.

- Integrated client with stateless, temporary address and prefix delegation

- A double-dereference in dhclient transmission of DHCPDECLINEs was

- Fix handling of format code 'Z'.

- Support "-1" argument in DHCPv6.

- Merge DHCPv6-only "dhcrelay6" into general-purpose "dhcrelay" (use
  "-6" option to select DHCPv6 mode).

- Fix handling of -A and -a flags in dhcrelay; it was failing to expand
  packet size as needed to add relay agent options.

- A bug in subnet6 parsing where options contained in subnet6 clauses would
  not be applied to clients addressed within that network was repaired.

- When configuring a "subnet {}" or "subnet6 {}" without an explicit
  shared-network enclosing it, the DHCP software would synthesize a
  shared-network to contain the subnet.  However, all configuration
  parameters within the subnet more intuitively belong "to any client
  on that interface", or rather the synthesized shared-network.  So,
  when a shared-network is synthesized, it is used to contain the
  configuration present inside the subnet {} clause.  This means that
  the configuration will be valid for all clients on that network, not
  just those addressed out of the stated subnet.  If you intended the
  opposite, the workaround is to explicitly configure an empty

- A bug was fixed where Information-Request processing was not sourcing
  configured option values.

- A warning was added since the DHCPv6 processing software does not yet
  support class statements.

- Compliation warnings on GCC 4.3 relating to bootp source address
  selection were repaired.

- The v6 BSD socket method was updated to use a single UDP BSD socket
  no matter how many interfaces are involved, differentiating the
  interfaces the packets were received on by the interface index supplied
  by the OS.

- The relay agent no longer listens to the All DHCP Servers Multicast

- A bug was fixed in data_string_sprintfa() where va_start was only called
  once for two invocations of vsprintf() variants.

- ERO (RFC 4994) server support.

- Basic and partial DHCPv6 leasequery support.

- Reliable DHCPv6 release (previous behavior, send release and exit, is
  still available with dhclient -6 -1 -r).

            Changes since 4.0.0 (new features)

- Added DHCPv6 rapid commit support.

- Added explicit parser support for zero-length DHCP options, such as
  rapid-commit, via format code 'Z'.

- It's now possible to update the "ends" field of a lease with OMAPI.
  This is useful if you want not only to release a lease, but also make
  it available for reuse right away.  Hat tip to Christof Chen.

- Fixed definition of the iaaddr hash functions to use the correct
  functions when referencing and dereferencing memory.

- Some definitions not in phase with the IANA registry were updated.

- Allocated interface IDs are better controlled ('u' bit set to zero,
  reserved IDs avoided).

- Unicast options are taken into account only for RENEWs.

- NoAddrsAvail answers to SOLICITs are always ADVERTISEs even when a SOLICIT
  carries a rapid-commit option.

- Return in place of raise an impossible condition when one tries to release
  an empty active lease.

- Timer granularity is now 1/100s in the DHCPv6 client.

- The dhclient-script was updated to create a host route for the default
  gateway if the supplied subnet mask for an IPv4 address was a /32.  This
  allows the client to work in 'captive' network environments, where the
  operator does not want clients to crosstalk directly.

- MINUS tokens should be parseable again.

- Multiple (up to "delayed-ack x;" maximum) DHCPv4 packets are now queued and
  released in bursts after single fsync() events when the upper limit is
  reached or if the receiving sockets go dry.  The practical upshot is
  that fsync-coupled server performance is now multiplicitively increased.
  The default delayed ack limit is 28.  Thanks entirely to a patch from
  Christof Chen.

            Changes since 4.0.0 (bug fixes)

- DHCP now builds on AIX.

- Exit with warning when DHCPv6-specific statements are used in the
  config file but -6 is not specified.

- Fixed "--version" flag in dhcrelay

- The 'min-secs' configuration parameter's log message has been updated to
  be more helpful.

- The warning logged when an address range doesn't fit in the subnets
  they were declared has been updated to be more helpful and identify the
  typo in configuration that created the spanning addresses.

- A bug in failover pool rebalancing that caused POOLREQ message ping-pongs
  was repaired.

- A flaw in failover pool rebalancing that could cause POOLREQ messages to
  be sent outside of the min-balance/max-balance scheduled intervals has
  been repaired.

- A cosmetic bug during potential-conflict recovery that caused the peer's
  'conflict-done' state message to be logged as 'unknown-state' has been
  repaired.  It is now logged correctly.

- A bug was fixed where the 'giaddr' may be used to find the client's subnet
  rather than its own 'ciaddr'.

- A log message was introduced to clarify the situation where a failover
  'address' parameter (the server's local address) did not resolve to an
  IPv4 address.

- The minimum site code value was set to 224 in 3.1.0 to track RFC3942.  This
  broke a lot of legacy site local configurations.  The new code in place will
  track site local space minimum option codes and logs a warning to encourage
  updates and exploration of site local code migration problems.  Option
  codes less than 128 in site local spaces remain inaccessible.

- A possible relay agent option bug was repaired where random server
  initialization state may have been used to signal the relay agent
  information options sub-option code for the 'END' of the option space.

- Fixes to allow code to compile and run on Solaris 9.

- Fixes to allow code to compile on Mac OS X Leopard (10.5).

- When server is configured with options that it overrides, a warning is
  issued when the configuration file is read, rather than at the time the
  option is overridden. This was important, because the warning was given
  every time the option was overridden, which could create a lot of
  unnecessary logging.

- Fixed a compilation problems on platforms that define a value for FDDI,
  which conflicts with a dhcp configuration syntax token by the same name.

- When a failover server suspects it has encountered a peer running a
  version 3.0.x failover server, a warning that the failover wire protocol
  is incompatible is printed.

- The failover server no longer issues a floating point error if it encounters
  a previously undefined option code.

- Fix startup error messages to report a missing "subnet6 declaration", rather
  than a missing "subnet declaration", when running as a DHCPv6 server.

- DHCPv6 client timestamp in DUID was based on the year 1970 rather
  than the year 2000.

- Warn when attempting to use a hardware parameter in DHCPv6.

- DHCPv6 released resources are now marked as released by the client.

- 'Soft' bindings have no more side-effects.

            Changes since 4.0.0b3

- The reverse dns name for PTR updates on IPv6 addresses has been fixed to
  use rather than default to and require user

- dhc6_lease_destroy() and dhc6_ia_destroy() now set lease and IA pointers
  to NULL after freeing, to prevent subsequent accesses to freed memory.

- The DHCPv6 server would not send the preference option unless the
  client requested it, via the ORO. This has been fixed, so the DHCPv6
  server will always send the preference value if it is configured.

- When addresses were passed as hints to the server in an IA, they were
  incorrectly handled, sometimes being treated as an error. Now the
  server will treat these as hints and ignore them if it cannot supply
  a requested address.

- If the client had multiple addresses, and one expired (was not renewed
  by the server), the client would continue to attempt to renew the same
  old address over and over.  Now, the client will omit any expired
  addresses from future Confirm, Renew, or Rebind messages.

- dhclient -6 will now select renew/rebind timers based upon the longest
  address expiration time rather than the shortest expiration time, in
  order to avoid cascading renewals in the event a server elects not to
  extend one of multiple IAADDR leases.

- The server now limits clients that request multiple addresses to one
  address per IA by default, which can be adjusted through the
  "limit-addrs-per-ia" configuration option.

- The DHCPv6 client now issues fresh transaction IDs on Renew and Rebind
  message exchanges, rather than using the most recent ID.

- The DHCPv6 server now replies to Information-Request messages.

- A bug was fixed in the dhclient-script for BSDs to correctly carry error
  codes through some conditions.

- The parsing of some options in the dhclient lease file, in particular
  the success DHCPv6 status-code, was fixed.

- A bug was fixed that caused the DHCPv6 ORO option to be corrupted with
  seemingly random values.

- A reference overleak in DHCPv6 shared network processing was repaired.

- ./configure now autodetects local database locations rather than trying
  to put dhcpd.leases and dhclient.leases in /usr/local/var/db, which no
  one ever has.

- Regression fix for bug where server advertised a IPv6 address in
  response to a SOLICIT but would not return the address in response
  to a REQUEST.

- A bug was fixed where the DHCPv6 server puts the NoAddrsAvail status
  code in the IA_NA was fixed.  The status code now appears in the root

            Changes since 4.0.0b2

- Clarified error message when lease limit exceeded

- Relative time may now be used as a qualifier for 'allow' and 'deny' access
  control lists.  These directives may be used to assist in re-addressing
  address pools without having to constantly reconfigure the server.  Please
  see 'man dhcpd.conf' for more information on allow/deny 'after time' syntax.
  Thanks to a patch from Christof Chen.

- The server will now include multiple IA_NA's and multiple IAADDRs within
  them, if advertised by the client.  It still only seeks to allocate one
  new address.

            Changes since 4.0.0b1

- Use different paths for PID and lease files when running in DHCPv4
  or DHCPv6 mode, so that servers for both protcols can be run
  simultaneously on a single interface.

- Fixed a buffer overflow error which could have allowed a denial
  of service under unusual server configurations

- Eliminated a spurious error message from the client

- A number of bugs with the internal handling of lease state on the
  server have been fixed. Some of these could cause server crashes.

- The peer_wants_leases() changes pulled up from 3.1.0 were corrected,
  'never used' leases will no longer consistently shift between servers
  on every pool rebalance run.

- sendmsg()/recvmsg() control buffers are now declared in such a way to
  ensure they are correctly aligned on all (esp. 64-bit) architectures.

- The client leasing subsystem was streamlined and corrected to account
  more closely for changes in client link attachment selection.

            Changes since 4.0.0a3

- The DHCP server no longer requires a "ddns-update-style" statement,
  and now defaults to "none", which means DNS updates are disabled.

- Log messages when failover peer names mismatch have been improved to
  point out the problem.

- Bug where server advertised a IPv6 address in response to a SOLICIT
  but would not return the address in response to a REQUEST. Thanks to
  Dennis Kou for finding the bug.

- Fixed an error causing the server to lock up on lease expiration,
  reported independently by Jothilingam Vasu and Dennis Kou.

- Fixed a ./configure bug where compile tests were failing due to
  "-Werror" (unused variable) rather than the actual test failure.  Lead
  to inconsistent and unworkable auto-configurations.

- Compilation with DLPI and -Werror has been repaired.

- Error in decoding IA_NA option if multiple interfaces are present
  fixed by Marcus Goller.

- DHCPv6 server Confirm message processing has been enhanced - it no
  longer replies only to clients with host {} records, it now replies
  as directed in RFC3315 section 18.2.2 - that is, to all clients
  regardless of the existence of bindings.

- A core dump during expired lease cleanup has been repaired.

- DDNS updates state information are now stored in 'binding scopes' that
  follow the leases through their lifecycles.  This enables DDNS teardowns
  on leases that are assigned and expired inbetween a server restart (the
  state is recovered from dhcpd.leases).  Arbitrary user-specified binding
  scopes ('set var = "value";') are not yet supported.

- Additional compilation problems on HP/UX have been repaired.

            Changes since 4.0.0a2

- Fix for startup where there are no IPv4 addresses on an interface.
  Thanks to Marcus Goller for reporting the bug.

- Fixed file descriptor leak on listen failure. Thanks to Tom Clark.

- Bug in server configuration parser caused server to get stuck on
  startup for certain bad pool declarations. Thanks to Guillaume
  Knispel for the bug report and fix.

- Code cleaned to remove warnings reported by "gcc -Wall".

- DHCPv6 is now the default. You can disable DHCPv6 support using the
  "--disable-dhcpv6" flag when you run the configure script.

- An internal database inconsistency bug was repaired where the server
  would segfault if a client attempted to renew a lease that had been
  loaded from persistent storage.

- 'request' and 'also request' syntaxes have been added to accommodate
  the DHCPv6 client configuration.  'send dhcp6.oro' is no longer

- Bug fixed where configuration file parsing did not work with
  zero-length options; this made it impossible to set the
  rapid-commit option.

- Bogus messages about host records with IPv4 fixed-addresses being of
  non-128-bits in length were removed.

            Changes since 4.0.0a1

- Bug in octal parsing fixed. Thanks to Bernd Fuhrmann for the report
  and fix.

- Autoconf now supplies proper flags for Solaris DHCPv6 builds.

- Fix for parsing error on some IPv6 addresses.

- Invalid CIDR representation for IPv6 subnets or ranges now checked
  for when loading configuration.

- Compilation on HP/UX has been repaired.  The changes should generally
  apply to any architecture that supplies SIOCGLIFCONF but does not
  use 'struct lifconf' structures to pass values.

- Two new operators, ~= and ~~, have been integrated to implement
  boolean matches by regular expression (such as may be used in
  class matching statements).  Thanks to a patch by Alexandr S.
  Agranovsky, which underwent slight modification.

- Fix for icmp packets on 64-bit systems (bug introduced in 4.0).

- A bug was fixed in interface discovery wherein an error identifying
  a server-configured interface with no IPv4 addresses would SEGV.

- Fixed a bug in which write_lease() might report a failure incorrectly

- Added support for DHCPv6 Release messages

- Added -x option to dhclient, which triggers dhclient processes
  to exit gracefully without releasing leases first

- All binaries (client, server, relay) now change directories
  to / before going into daemon mode, so as not to hold $CWD open

- Fixed a bug parsing DHCPv6 client-id's in host-identifier statements

- Fixed a bug with the 'ddns-updates' boolean server configuration
  parameter, which caused the server to fail.

            Changes since 4.0.0-20070413

- Old (expired) leases are now cleaned.

- IPv6 subnets now have support for arbitrary allocation ranges via
  a new 'range6' configuration directive.

- An obviated option code hash lookup to find D6O_CLIENTID was removed.

- Corrected some situations where variables might be used without being

- Silenced several other compiler warnings.

- Include the more standard sys/uio.h rather than rely upon other
  header files to include it (fixes a BSD 4.2 compile failure).

- Duplicate dhclient-script updates for DHCPv6 to all provided scripts.

- DHCPv4 I/O methods that failed to sense hardware address were corrected.

- DHCPv4 is now the default (as documented) rather than DHCPv6.  The default
  was set to DHCPv6 to facilitate ease early development, and forgotten.

- Corrected a segmentation violation in DHCPv4 socket processing.

- dhclient will now fork() into the background once it binds to an
  IPv6 address, or immediately if the -n flag is supplied.

- -q is now the default behaviour on dhclient, with -d or -v enabling
  non-quiet (stderr logging) mode.

- Fix documentation of the domain-search atom (quoted, with commas).

- Document DHCPv6 options presently in the default table.

- Replaced ./configure shellscripting with GNU Autoconf.

             Changes since 3.1.0 (NEW FEATURES)

- DHCPv6 Client and Server protocol support.  Use '-6' to run the daemons
  as v6-only.  Use '-4' to run the daemons as v4-only (default.  There is
  no support currently for both.

- Server support for multiple IA_NA options, containing at most one
  IAADDR option.

- Client support for one IA_NA option, containing any number of IAADDR

- Server support for the DHCPv6 Information-request message.

- Inappropriate unicast DHCPv6 messages sent to the server are now
  discarded, and this has rearchitected the IO system slightly.

- The DHCPv6 server DUID defaults to type 1, is persistently stored in
  the leases database, and can be over-ridden (either completely, or by
  specifying type 1 or type 2).

- The server only uses Rapid-Commit if it has been configured with the
  Rapid-Commit option and the client requests it.

- DDNS support. We now update AAAA records in the same place we would
  update A records, if we have an IPv6 address. We also generate IP6.ARPA
  style names for PTR records if we're dealing with an IPv6 address.  Both
  A and AAAA updates are done using the same 'fqdn.' virtual option space
  (although the DHCPv4 FQDN and DHCPv6 FQDN options are formatted
  differently, they both use the same code here).

- The Linux dhclient-script attempts to set and remove assigned addresses,
  and to configure /etc/resolv.conf from nameserver and domain name
  configurations.  It can be extended to configure other parameters.

- Initial DHCPv6 lease support.

- The IO system now tracks all local IP addresses, so that the DHCP
  applications (particularly the dhcrelay) can discern between what frames
  were transmitted to it, and what frames are being carried through it which
  it should not intercept.

            Changes since 3.1.0 (Maintenance)

- A bug was repaired where MAC Address Affinity for virgin leases always
  mapped to the primary.  Virgin leases now have an interleaved preference
  between primary and secondary.

- A bug was repaired where MAC Address Affinity for clients with no client
  identifier was sometimes mishashed to the peer.  Load balancing during
  runtime and pool rebalancing were opposing.

- An assertion in lease counting relating to reserved leases was repaired.

- The subnet-mask option inclusion now conforms with RFC2132 section 3.3;
  it will only appear prior to the routers option if it is present on the
  Parameter-Request-List.  The subnet-mask option will also only be
  included by default (if it is not on the PRL) in response to DISCOVER
  or REQUEST messages.

- The FQDN option is only supplied if the client supplied an FQDN option or
  if the FQDN option was explicitly requested on the PRL.

- Dynamic BOOTP leases are now load balanced in failover.

            Changes since 3.1.0rc1

- The parse warning that 'deny dyanmic bootp;' must be configured for
  failover protected subnets was removed.

            Changes since 3.1.0b2

- Failover rebalance events no longer play ping pong with round errors
  (moving leases between free and back to backup where there are an
  odd number of leases).

- The 'pool' log line has been split into two messages, one before the
  rebalance run, and one after.

- Any queued BNDACKs are transmitted before transmitting new BNDUPDs.
  This enforces the correct sequence of events for the remote server
  processing these messages.

            Changes since 3.1.0b1

- Fixed a bug that caused OMAPI clients to freeze when opening lease

- A new server config option "fqdn-reply" specifies whether the server
  should send out option 81 (FQDN).  Defaults to "on".  If set to "off",
  the FQDN option is not sent, even if the client requested it.  This is
  needed because some clients misbehave otherwise.  Thanks to Christof Chen
  at Allianz.

- Allow trace output files (-tf option) to be overwritten, rather than
  crashing dhcpd if the file already exists

- A bug was fixed that caused dhcpd to segfault if a pool was declared
  outside the scope of a subnet in dhcpd.conf.

- Some uninitialized values were repaired in dhcpleasequery.c that
  caused the server to abort.

- A new server config option, 'do-reverse-updates', has been added
  which causes the server to abstain from performing updates on PTR
  records.  Thanks to a patch from Christof Chen at Allianz.

- A bug was repaired in subencapsulation support, where spaces separated
  by empty spaces would not get included.

- A bug in dhclient was repaired which caused it to send parameter request
  lists of 55 bytes in length no matter how long the declared PRL was.

- 'dhcp.c(3953): non-null pointer' has been repaired.  This fixes a flaw
  wherein the DHCPv4 server may ignore a configured server-identifier.

- A flaw in failover startup sequences was repaired that sometimes left
  the primary DHCP server's pool rebalance schedules unscheduled.

- Corrected a flaw that broke encapsulated spaces included due to presence
  on the parameter request list.

            Changes since 3.1.0a3

- Some spelling fixes.

            Changes since 3.1.0a2

- A bug was fixed where attempting to permit leasequeries results in a
  fatal internal error, "Unable to find server option 49".

- A bug was fixed in dhclient rendering the textual output form of the
  domain-search option syntax.

            Changes since 3.1.0a1

- A bug in the FQDN universe that added FQDN codes to the NWIP universe's
  hash table was repaired.

- The servers now try harder to transmit pending binding updates when
  entering normal state.

- UPDREQ/UPDREQALL handling was optimized - it no longer dequeues and
  requeues all pending updates.  This should reduce the number of spurious
  'xid mismatch' log messages.

- An option definition referencing leak was fixed, which resulted in early
  termination of dhclient upon the renewal event.

- Some default hash table sizes were tweaked, some upwards, some downwards.
  3.1.0a1's tables resulted in a reduction in default server memory use.
  The new selected values provide more of a zero sum (increasing the size
  of tables likely to be populated, decreasing the size of tables unlikely).

- Lease structures appear in three separate hashes: by IP address, by UID,
  and by hardware address.  One type of table was used for all three, and
  improvements to IP address hashing were applied to all three (so UID and
  hardware addresses were treated like 4-byte integers).  There are now two
  types of tables, and the uid/hw hashes use functions more appropriate
  to their needs.

- The max-lease-misbalance percentage no longer causes scheduled rebalance
  runs to be skipped: it still governs the schedule, but every scheduled
  run will attempt balance.

- A segfault bug in recursive encapsulation support has been corrected.

            Changes since 3.0 (New Features)

- A workaround for certain STSN servers that send a mangled domain-name
  option was introduced for dhclient.  The client will now accept corrupted
  server responses, if they contain a valid DHCP_MESSAGE_TYPE (OFFER, ACK,
  or NAK).  The server will continue to not accept corrupt client packets.

- Support for 'reserved' (pseudo-static) and BOOTP leases via failover
  was introduced.

- Support for adding, removing, and managing class and subclass statements
  via OMAPI.

- The failover implementation was updated to comply with revision 12 of
  the protocol draft.

- 'make install' now creates the initial zero-length dhcpd.leases file if
  one does not already exist on the system.

- RFC3942 compliance, site-local option spaces start at 224 now, not 128.

- The Load Balance Algorithm was misimplemented.  The current implementation
  matches RFC 3074.

- lcase() and ucase() configuration expressions have been added which adjust
  their arguments from upper to lower and lower to upper cases respectively.
  Thanks to a patch from Albert Herranz.

- The dhclient 'reject ...;' statement, which rejects leases given by named
  server-identifiers, now permits address ranges to be specified in CIDR
  notation.  Thanks to a patch from David Boyce.

- The subnet-mask option is now supplied by default, but at lowest
  priority.  This helps a small minority of clients that provide parameter
  request lists, but do not list the subnet-mask option because they were
  designed to interoperate with a server that behaves in this manner.

- The FQDN option is similarly supplied even if it does not appear on the
  parameter request list, but not to the exclusion of options that do
  appear at the parameter request list.  Up until now it had ultimate
  priority over the client's parameter request list.

- Varying option space code and length bit widths (8/16/32) are now
  supported.  This is a milestone in achieving RFC 3925 "VIVSO" and
  DHCPv6 support.

- A new common (server or client) option, 'db-time-format local;', has
  been added which prints the local time in /var/db/dhcpd.leases rather
  than UTC.  Thanks to a patch from Ken Lalonde.

- Some patches to improve DHCP Server startup speed from Andrew Matheson
  have been incorporated.

- Failover pairs now implement 'MAC Affinity' on leases moving from the
  active to free states.  Leases that belonged to the failover secondary
  are moved to BACKUP state rather than FREE upon exiting EXPIRED state.
  If lease rebalancing must move leases, it tries first to move leases
  that belong to the peer in need.

- The server no longer sends POOLREQ messages unless the pool is severely
  misbalanced in the peer's favor (see 'man dhcpd.conf' for more details).

- Pool rebalance events no longer happen upon successfully allocating a
  lease.  Instead, they happen on a schedule.  See 'man dhcpd.conf' for the
  min-balance and max-balance statements for more information.

- The DHCP Relay Agent Information Option / Link Selection Sub-Option
  is now supported.  (See RFC3527 for details).

- A new DDNS related server option, update-conflict-detection, has been
  added.  If this option is enabled, dhcpd will perform normal DHCID
  conflict resolution (the default).  If this option is disabled, it will
  instead trust the assigned name implicitly (removing any other bindings
  on that name).  This option has not been made available in dhclient.

- In those cases where the DHCP software manufactures an IP header (to
  transmit via bpf, lpf, etc), the IP TTL the software selects has been
  increased from 16 to 128.  This is intended to match Microsoft Windows
  DHCP Client behaviour, to increase compatibility.

- 'ignore client-updates;' now has behaviour that is different from
  'deny client-updates;'.  The client's request is not truly ignored,
  rather it is encouraged.  Should this value be configured, the server
  updates DNS as though client-updates were set to 'deny'.  That is, it
  enters into DNS whatever it is configured to do already, provided it is
  configured to.  Then it sends a response to the client that lets the
  client believe it is performing client updates (which it will), probably
  for a different name.  In essence, this lets the client do as it will,
  ignoring this aspect of their request.

- Support for compressed 'domain name list' style DHCP option contents, and
  in particular the domain search option (#119) was added.

- The DHCP LEASEQUERY protocol as defined in RFC4388 is now implemented.
  LEASEQUERY lets you query the DHCP server for information about a lease,
  using either an IP address, MAC address, or client identifier.  Thanks
  to a patch from Justin Haddad.

- DHCPD is now RFC2131 section 4.1 compliant (broadcast to all-ones ip and
  ethernet mac address) on the SCO platform specifically without any strange
  ifconfig hacks.  Many thanks go to the Kroger Co. for donating the
  hardware and funding the development.

- A new common configuration executable statement, execute(), has been
  added.  This permits dhcpd or dhclient to execute a named external
  program with command line arguments specified from other configuration
  language.  Thanks to a patch written by Mattias Ronnblom, gotten to us
  via Robin Breathe.

- A new dhcp server option 'adaptive-lease-time-threshold' has been added
  which causes the server to substantially reduce lease-times if there are
  few (configured percentage) remaining leases.  Thanks to a patch submitted
  from Christof Chen.

- Encapsulated option spaces within encapsulated option spaces is now
  formally supported.

            Changes since 3.0.6rc1

- supersede_lease() now requeues leases in their respective hardware
  address hash bucket.  This mirrors client identifier behaviour.

            Changes since 3.0.5

- Assorted fixes for broken network devices:  Packet length is now
  determined from the IP header length field to finally calculate the
  UDP payload length, because some NIC drivers return more data than
  they actually received.

- UDP packets are now stored in aligned data structures.

- A logic error in omapi interface code was repaired that might result in
  incorrectly indicating 'up' state when any flags were set, rather than
  specifically the INTERFACE_REQUESTED flag.  Thanks to a patch from
  Jochen Voss which got to us via Andrew Pollock at Debian.

- A reference leak on binding scopes set by ddns updates was repaired.

- A memory leak in the minires_nsendsigned() function call was repaired.
  Effectively, this leaked ~176 bytes per DDNS update.

- In the case where an "L2" DHCP Relay Agent (one that does not set giaddr)
  was directly attached to the same broadcast domain as the DHCP server,
  the RFC3046 relay agent information option was not being returned to the
  relay in the server's replies.  This was fixed; the dhcp server no longer
  requires the giaddr to reply with relay agent information.  Note that
  this also improves compatibility with L2 devices that "intercept" DHCP
  packets and expect relay agent information even in unicast (renewal)
  replies.  Thanks to a patch from Pekka Silvonen.

- A bug was fixed where the BOOTP header 'sname' field had a value, the
  copy written to persistent storage was actually the contents of the
  'file' field.

- A bug was fixed where the nwip virtual option space was referencing
  the fqdn option's virtual option space's option cache.

- Timestamp parsing errors that indicated missing "minutes" fields rather
  than the actually missing "seconds" fields have been repaired thanks to
  a patch from Kevin Steves.

- A grammar error in the dhclient.8 manpage was repaired thanks to a patch
  from Chris Wagner.

- Several spelling typos were repaired, and some cross-references to other
  relevant documents were included in the manpages, thanks to a patch
  by Andrew Pollock which got to us via Tomas Pospisek.

- Some bugs were fixed in the 'emergency relay agent options hologram'
  which is used to retain relay agent option contents from when the
  client was in INIT or REBIND states.  This should solve problems where
  relay agent options were not echoed from the server, even when giaddr
  was set.

- dhclient now closes its descriptor to dhclient.leases prior to executing
  dhclient-script.  Thanks to a patch from Tomas Pospisek.

- The server's "by client-id" and "by hardware address" hash table lists
  are now sorted according to the preference to re-allocate that lease to
  returning clients.  This should eliminate pool starvation problems
  arising when "INIT" clients were given new leases rather than presently
  active ones.

            Changes since 3.0.5rc1

- A bug was repaired in fixes to the dhclient, which sought to run the
  dhclient-script with the 'EXPIRE' state should it receive a NAK in
  response to a REQUEST.  The client now iterates the PREINIT state
  after the EXPIRE state, so that interfaces that might be configured
  'down' can be brought back 'up' and initialized.

- DHCPINFORM handling for clients that properly set ciaddr and come to the
  server via a relay aget has been repaired.

            Changes since 3.0.4

- A warning that host statements declared within subnet or shared-network
  scopes are actually global has been added.

- The default minimum lease time (if min-lease-time was not specified)
  was raised from 0 to 300.  0 is not thought to be sensible, and is
  known to be damaging.

- Added additional fatal error sanity checks surrounding lease binding
  state count calculations (free/active counts used for failover pool

- Some time value size fixes in 3.0.4 brought on from FreeBSD /usr/ports were
  misapplied to server values rather than client values.  The server no longer
  advertises 8-byte lease-time options when on 64-bit platforms.

- A bug where leases not in ACTIVE state would get billed to billed classes
  (classes with lease limitations) was fixed.  Non-active leases OFFERed
  to clients are no longer billed (but billing is checked before offering).

- The dhcpd.conf.5 manpage was updated in regard to the ddns-domainname
  configuration option - the default configuration and results should be
  more clear now.

- If the dhclient were to receive a DHCPNAK while it was in the RENEW
  state (and consequently, had an active, 'bound' address and related
  configuration options), it would fail to 'tear down' this information
  before proceeding into INIT state.  dhclient now iterates the dhclient-
  script with the 'EXPIRE' action to cause these teardowns prior to entering
  INIT state.  Thanks to a patch from Chris Zimmerman.

- The omapi.1 manpage had some formatting errors repaired thanks to a patch
  from Yoshihiko Sarumaru.

- A few lines of code that were failover-specific were moved within
  #if defined() clauses so that compilation without failover could be
  made possible.

- The log message emitted when the 'leased-address' value was not available
  in dhcpd.conf "executable statements" has been updated to be more helpful.
  Manpage information for this value has also been updated.

- Abandoned or dissociated (err condition) leases now remove any related
  dynamic dns bindings.  Thanks to a patch from Patrick Schoo.

- Attempting to write a new lease file to replace a corrupt (due to
  encountering non-retryable errors during writing) lease file should
  no longer result in an infinite recursion.

- Host declaration hardware addresses and client identifiers may only be
  configured once.  dhcpd will now fail to load config files that specify
  multiple identifiers (previous versions would silently over-ride the
  value with the later configured value).

- Several option codes that have been allocated since our last release
  have been named and documented.

- Option names of the form "unknown-123" have been removed from the in-
  memory hash tables.  In order to support options of these names that
  may appear in dhclient.leases or similar in previous versions, the
  parser will now find the new option code definition, or mock up a
  generic option code definition.  This should result in a smooth
  transition from one name to the other, as the new name is used to
  write new output.

            Changes since 3.0.4rc1

- The dhcp-options.5 manpage was updated to correct indentation errors
  thanks to a patch from Jean Delvare.

            Changes since 3.0.4b3

- Some manual pages were clarified pursuant to discussion on the dhcp-server
  mailing list.

            Changes since 3.0.4b2

- Null-termination sensing for certain clients that unfortunately require
  it in DHCPINFORM processing was repaired.

- The host-name option and a few others were moved from "X" format to "t"
  format to be compatible with new NULL handling functions.

- DHCPINFORM processing is a little more careful about return addressing
  its responses, or if responding via a relay.  The INFORM related
  messages also log the 'effective client ip address' rather than the
  client's supplied ciaddr (since some clients produce null ciaddrs).

- The server was inappropriately sending leases to the RESET state in the
  event that multiple active leases were found to match a singly-identified
  client.  This was changed to RELEASED (by accepting a different, ACTIVE
  binding, the client is implicitly releasing its lease).  This repairs a
  bug wherein secondary servers in failover pairs detecting this condition
  move leases to RESET, and primaries refuse to accept that state
  transition (properly).

- The memset-after-dmalloc() changes made in 3.0.4b1 have been backed out.

            Changes since 3.0.4b1

- Command line parsing in omshell was repaired - it no longer closes
  STDIN after reading one line.

- The resolver library no longer closes the /etc/resolv.conf file
  descriptor it opened twice.

- Changes to trailing NULL removal in 't' option-atoms has been rethought,
  it now includes 'd' (domain name) types, and tries hard not to rewind an
  option beyond the start of the text field it is un-terminating.

            Changes since 3.0.3

- A DDNS update handling function was misusing the DNS error codes, rather
  than the internal generic result enumeration.  The result is a confusing
  syslog line, logging the wrong condition.

- The DHCP Server was not checking pool balance in the case where it brought
  a non-ACTIVE lease out of storage for a client that was returning to use
  a lease it once had long ago, and had since expired.

- Failover peers no longer bother to look for free leases to allocate when
  they already found the client's ACTIVE lease.  DISCOVERs are load balanced
  whether freely-allocated or not, unless the server doubts the peer has
  leases to allocate.

- Fixed a bug in dhcrelay agent addition code that suppressed trailing
  PAD options - it was suppressing only one trailing PAD option, rather
  than the entire block of them.

! Fixed some unlikely overlapping-region memcpy() bugs in dhcrelay agent
  option addition and stripping code.  Added a few sanity checks.  Although
  highly improbable, due to requiring the reception of a DHCP datagram well
  in excess of all known to be used physical MTU limitations, it is possible
  this may have been used in a stack overflow security vulnerability.  Thanks
  to a patch from infamous42md.

! Added some sanity checks to OMAPI connection/authentication code.
  Although highly improbable, due to having to deliver in excess of 2^32
  bytes of data via the OMAPI channel, not to mention requiring dhcpd to
  be able to malloc() a memory region 2^32 bytes in size, it was possible
  this might have resulted in a heap overflow security vulnerability.
  Thanks to a patch from infamous42md.

- dmalloc() memset()'s the non-debug (data) portion of the allocated
  memory to zero.  Code that memset()'s the result returned by dmalloc() to
  zero is redundant.  These redundancies were removed.

- Some type declaration corrections to u_int16_t were made in common/tr.c
  (Token Ring support) thanks to a patch from Jason Vas Dias at Red Hat.

- A failover bug that was allowing leases that EXPIRED or were RELEASED
  where tsfp and tstp are identical timestamps to languish in these
  transitional states has been repaired.  As a side effect, lease
  databases should be kept more consistent overall, not just for these
  transitional states.

- If the lease db is deleted out from under the daemon, and it moves to rewrite
  the db, it will go ahead with the operation and move the new db into place
  once it detects the old db does not exist.

- dhclient now ignores IRDA, SIT, and IEEE1394 network interfaces, as it
  is either nonsensical or (in the case of IEEE1394) is not known to support
  these interfaces.  Thanks to Marius Gedminas and Andrew Pollock of Debian.

- Some previously undocumented reasons for dhclient-script invoking has
  been documented in the dhclient-script.8 manpage.

- Failover potential expiry calculations (TSTP) have been corrected.  Results
  should be substantially more consistent, and proper given the constraints.

- Adjusted lease state validation checks in potential-conflict, to
  account for possible clock skew similarly to normal state, and several
  previously illegal transitions were made legal (ex: active->released).

- An impossible sanity check was removed from omapi/buffer.c, thanks to a
  patch from 'infamous42md'.

- An OMAPI host/network byte order problem in lease time values has been

- Several minor bugs, largely relating to treating 8-byte time values as
  4-byte entities, have been repaired after careful review of the FreeBSD
  ports collection's patch set.  Thanks to the nameless entities who have
  contributed to the FreeBSD ports.

- When writing a trace file, the file is now created with permissions 0600,
  to help administrators avoid accidentally publicising sensitive config

- The calculation of the maximum size of DHCP packets no longer includes
  Ethernet framing overhead.  The result is that the 'Maximum Message
  Size' option advertised by clients, or the default value 576, is no
  longer reduced by 14 bytes, and instead directly reflects the IP level
  MTU (and the default, minimum allowed IP MTU of 576).

- The special status of RELEASED/EXPIRED/RESET leases when a server
  is operating in partner-down was fixed.  It no longer requires a
  lease be twice the MCLT beyond STOS to 'reallocate', and the expiry
  event to turn these into FREE leases without peer acknowledgement
  (after STOS+MCLT) has been repaired.

- Compilation on older Solaris systems (lacking /usr/include/sys/int_types.h)
  has been repaired.

- "append"ing a string onto the end of a "t" type option (such as the
  domain-name field) that had been improperly NULL-terminated by the
  DHCP server will no longer result in a truncated string containing
  only the option from the server, and not the expected appended value.
  Thanks to a patch from Jason Vas Dias at Red Hat.

- File handlers on configuration state (config files and lease dbs) should
  be treated consistently, regardless of whether TRACING is defined or not.

- The Linux build environment has had some minor improvements - better
  sensing of 64-bit pointer sizes (only used for establishing an icmp_id),
  and corrections to #if operators regarding LINUX_MAJOR should it ever
  move to 3.[01].x.

- The server now tries harder to survive the condition where it is unable
  to open a new lease file to rewrite the lease state database.

            Changes since 3.0.3b3

- dhclient.conf documentation for interface {} was updated to reflect recent
  discussion on the dhcp-hackers mailing list.

- In response to reports that the software does not compile on GCC 4.0.0,
  -Werror was removed from Makefile.conf for all platforms that used it.
  We will address the true problem in a future release; this is a temporary

            Changes since 3.0.3b2

- An error in code changes introduced in 3.0.3b2 was corrected, which caused
  static BOOTP clients to receive random addresses.

            Changes since 3.0.3b1

- A bug was fixed in BOOTPREQUEST handling code wherein stale references to
  host records would be left behind on leases that were not allocated to the
  client currently booting (eg in the case where the host was denied booting).

- The dhcpd.conf.5 manpage was updated to be more clear in regards to
  multiple host declarations (thanks to Vincent McIntyre).  'Interim' style
  dynamic updates were also retouched.

            Changes since 3.0.2

- A bug was fixed where a server might load balance a DHCP REQUEST to its
  peer after already choosing not to load balance the preceding DISCOVER.
  The peer cannot allocate the originating server's lease.

- In the case where a secondary server lost its stable storage while the
  primary was still in communications-interrupted, and came back online,
  the lease databases would not be fully transferred to the secondary.
  This was due to the secondary errantly sending an extra UPDREQ message
  when the primary made its state transition to PARTNER-DOWN known.

- The package will now compile cleanly in gcc 3.3 and 3.4.  As a side effect,
  lease structures will be 9 bytes smaller on all platforms.  Thanks to
  Jason Vas Dias at Red Hat.

- Interface discovery code in DISCOVER_UNCONFIGURED mode is now
  properly restricted to only detecting broadcast interfaces.  Thanks
  to a patch from Jason Vas Dias at Red Hat.

- decode_udp_ip_header was changed so that the IP address was copied out
  to a variable, rather than referenced by a pointer.  This enforces 4-byte
  alignment of the 32-bit IP address value.  Thanks to a patch from Dr.
  Peter Poeml.

- An incorrect log message was corrected thanks to a patch from
  Dr. Peter Poeml.

- A bug in DDNS was repaired, where if the server's first DDNS action was
  a DDNS removal rather than a DDNS update, the resolver library's
  retransmit timer and retry timer was set to the default, implying a
  15 second timeout interval.  Which is a little excessive in a synchronous,
  single-threaded system.  In all cases, ISC DHCP should now hold fast to
  a 1-second timeout, trying only once.

- The siaddr field was being improperly set to the server-identifier when
  responding to DHCP messages.  RFC2131 clarified the siaddr field as
  meaning the 'next server in the bootstrap process', eg a tftp server.
  The siaddr field is now left zeroed unless next-server is configured.

- mockup_lease() could have returned in an error condition (or in the
  condition where no fixed-address was found matching the shared
  network) with stale references to a host record.  This is probably not
  a memory leak since host records generally never die anyway.

- A bug was repaired where failover servers would let stale client identifiers
  persist on leases that were reallocated to new clients not sending an id.

- Binding scopes ("set var = value;") are now removed from leases allocated
  by failover peers if the lease had expired.  This should help reduce the
  number of stale binding scopes on leases.

- A small memory leak was closed involving client identifiers larger than
  7 bytes, and failover.

- Configuring a subnet in dhcpd.conf with a subnet mask of 32 bits might
  cause an internal function to overflow heap.  Thanks to Jason Vas Dias
  at Red Hat.

- Some inconsistencies in treating numbers that the lexer parsed as 'NUMBER'
  or 'NUMBER_OR_NAME' was repaired.  Hexadecimal parsing is affected, and
  should work better.

- In several cases, parse warnings were being issued before the lexical
  token had been advanced to the token whose value was causing an error...
  causing parse warnings to claim the problem is on the wrong token.

- Host declarations matching on client identifier for dynamic leases will
  no longer match fixed-address host declarations (this is now identical
  to behaviour for host records matching on hardware address).

            Changes since 3.0.2rc3

- A previously undocumented configuration directive, 'local-address',
  was documented in the dhcpd.conf manpage.

            Changes since 3.0.2rc2

- Two variables introduced in 3.0.2b1 were used without being initialized
  in the case where neither the FILE nor SNAME fields were available for
  overloading.  This was repaired.

- A heretofore believed to be impossible corner case of the option
  overloading implementation turned out to be possible ("Unable to sort
  overloaded options after 10 tries.").  The implementation was reworked
  to consider the case of an option so large it would require more than
  three chunks to fit.

- Many other instances of variables being used without being initialized
  were repaired.

- An uninitialized variable in omapi_io_destroy() led to the discovery
  that this function may result in orphaned pointers (and hence, a memory

            Changes since 3.0.2rc1

- allocate_lease() was rewritten to repair a bug in which the server would
  try to allocate an ABANDONED lease when FREE leases were available.

            Changes since 3.0.2b1

- Some dhcp-eval.5 manpage formatting was repaired.

            Changes since 3.0.1

- A bug was fixed in the server's 'option overloading' implementation,
  where options loaded into the 'file' and 'sname' packet fields were
  not aligned precisely as rfc2131 dictates.

© 2001-2018 Internet Systems Consortium

For assistance with problems and questions for which you have not been able to find an answer in our Knowledge Base, we recommend searching our community mailing list archives and/or posting your question there (you will need to register there first for your posts to be accepted). The bind-users and the dhcp-users lists particularly have a long-standing and active membership.

ISC relies on the financial support of the community to fund the development of its open source software products. If you would like to support future product evolution and maintenance as well having peace of mind knowing that our team of experts are poised to provide you with individual technical assistance whenever you call upon them, then please consider our Professional Subscription Support services - details can be found on our main website.

  • There is no feedback for this article
Quick Jump Menu