Network Working Group                                          E. Gavron
Request for Comments: 1535                            ACES Research Inc.
Category: Informational                                     October 1993


             A Security Problem and Proposed Correction
                  With Widely Deployed DNS Software

Status of this Memo

  This memo provides information for the Internet community.  It does
  not specify an Internet standard.  Distribution of this memo is
  unlimited.

Abstract

  This document discusses a flaw in some of the currently distributed
  name resolver clients.  The flaw exposes a security weakness related
  to the search heuristic invoked by these same resolvers when users
  provide a partial domain name, and which is easy to exploit (although
  not by the masses).  This document points out the flaw, a case in
  point, and a solution.

Background

  Current Domain Name Server clients are designed to ease the burden of
  remembering IP dotted quad addresses.  As such they translate human-
  readable names into addresses and other resource records.  Part of
  the translation process includes understanding and dealing with
  hostnames that are not fully qualified domain names (FQDNs).

  An absolute "rooted" FQDN is of the format {name}{.} A non "rooted"
  domain name is of the format {name}

  A domain name may have many parts and typically these include the
  host, domain, and type.  Example:  foobar.company.com or
  fooschool.university.edu.

Flaw

  The problem with most widely distributed resolvers based on the BSD
  BIND resolver is that they attempt to resolve a partial name by
  processing a search list of partial domains to be added to portions
  of the specified host name until a DNS record is found.  This
  "feature" is disabled by default in the official BIND 4.9.2 release.

  Example: A TELNET attempt by    [email protected]
                            to    UnivHost.University.EDU



Gavron                                                          [Page 1]

RFC 1535               DNS Software Enhancements            October 1993


  The resolver client will realize that since "UnivHost.University.EDU"
  does not end with a ".", it is not an absolute "rooted" FQDN.  It
  will then try the following combinations until a resource record is
  found:

               UnivHost.University.EDU.Tech.ACES.COM.
               UnivHost.University.EDU.ACES.COM.
               UnivHost.University.EDU.COM.
               UnivHost.University.EDU.

Security Issue

  After registering the EDU.COM domain, it was discovered that an
  unliberal application of one wildcard CNAME record would cause *all*
  connects from any .COM site to any .EDU site to terminate at one
  target machine in the private edu.com sub-domain.

  Further, discussion reveals that specific hostnames registered in
  this private subdomain, or any similarly named subdomain may be used
  to spoof a host.

       Example:        harvard.edu.com.        CNAME   targethost

  Thus all connects to Harvard.edu from all .com sites would end up at
  targthost, a machine which could provide a Harvard.edu login banner.

  This is clearly unacceptable.  Further, it could only be made worse
  with domains like COM.EDU, MIL.GOV, GOV.COM, etc.

Public vs. Local Name Space Administration

  The specification of the Domain Name System and the software that
  implements it provides an undifferentiated hierarchy which permits
  delegation of administration for subordinate portions of the name
  space.  Actual administration of the name space is divided between
  "public" and "local" portions.  Public administration pertains to all
  top-level domains, such as .COM and .EDU.  For some domains, it also
  pertains to some number of sub-domain levels.  The multi-level nature
  of the public administration is most evident for top-level domains
  for countries.  For example in the Fully Qualified Domain Name,
  dbc.mtview.ca.us., the portion "mtview.ca.us" represents three levels
  of public administration.  Only the left-most portion is subject to
  local administration.








Gavron                                                          [Page 2]

RFC 1535               DNS Software Enhancements            October 1993


  The danger of the heuristic search common in current practise is that
  it it is possible to "intercept" the search by matching against an
  unintended value while walking up the search list.  While this is
  potentially dangerous at any level, it is entirely unacceptable when
  the error impacts users outside of a local administration.

  When attempting to resolve a partial domain name, DNS resolvers use
  the Domain Name of the searching host for deriving the search list.
  Existing DNS resolvers do not distinguish the portion of that name
  which is in the locally administered scope from the part that is
  publically administered.

Solution(s)

  At a minimum, DNS resolvers must honor the BOUNDARY between local and
  public administration, by limiting any search lists to locally-
  administered portions of the Domain Name space.  This requires a
  parameter which shows the scope of the name space controlled by the
  local administrator.

  This would permit progressive searches from the most qualified to
  less qualified up through the locally controlled domain, but not
  beyond.

  For example, if the local user were trying to reach:

       [email protected] from
       starburst,astro.DESERTU.EDU,

  it is reasonable to permit the user to enter just chief.admin, and
  for the search to cover:

       chief.admin.astro.DESERTU.EDU
       chief.admin.DESERTU.EDU

  but not

       chief.admin.EDU

  In this case, the value of "search" should be set to "DESERTU.EDU"
  because that's the scope of the name space controlled by the local
  DNS administrator.

  This is more than a mere optimization hack.  The local administrator
  has control over the assignment of names within the locally
  administered domain, so the administrator can make sure that
  abbreviations result in the right thing.  Outside of the local
  control, users are necessarily at risk.



Gavron                                                          [Page 3]

RFC 1535               DNS Software Enhancements            October 1993


  A more stringent mechanism is implemented in BIND 4.9.2, to respond
  to this problem:

  The DNS Name resolver clients narrows its IMPLICIT search list IF ANY
  to only try the first and the last of the examples shown.

  Any additional search alternatives must be configured into the
  resolver EXPLICITLY.

  DNS Name resolver software SHOULD NOT use implicit search lists in
  attempts to resolve partial names into absolute FQDNs other than the
  hosts's immediate parent domain.

  Resolvers which continue to use implicit search lists MUST limit
  their scope to locally administered sub-domains.

  DNS Name resolver software SHOULD NOT come pre-configured with
  explicit search lists that perpetuate this problem.

  Further, in any event where a "." exists in a specified name it
  should be assumed to be a fully qualified domain name (FQDN) and
  SHOULD be tried as a rooted name first.

  Example:  Given  [email protected] connecting to e.f.g.h  only two tries
            should be attempted as a result of using an implicit
            search list:

               e.f.g.h.  and e.f.g.h.b.c.d.

            Given [email protected]. connecting to host those same two
            tries would appear as:

               x.b.c.d.  and x.

  Some organizations make regular use of multi-part, partially
  qualified Domain Names.  For example, host foo.loc1.org.city.state.us
  might be used to making references to bar.loc2, or mumble.loc3, all
  of which refer to whatever.locN.org.city.state.us

  The stringent implicit search rules for BIND 4.9.2 will now cause
  these searches to fail.  To return the ability for them to succeed,
  configuration of the client resolvers must be changed to include an
  explicit search rule for org.city.state.us.  That is, it must contain
  an explicit rule for any -- and each -- portion of the locally-
  administered sub-domain that it wishes to have as part of the search
  list.





Gavron                                                          [Page 4]

RFC 1535               DNS Software Enhancements            October 1993


References

  [1] Mockapetris, P., "Domain Names Concepts and Facilities", STD 13,
      RFC 1034, USC/Information Sciences Institute, November 1987.

  [2] Mockapetris, P., "Domain Names Implementation and Specification",
      STD 13, RFC 1035, USC/Information Sciences Institute, November
      1987.

  [3] Partridge, C., "Mail Routing and the Domain System", STD 14, RFC
      974, CSNET CIC BBN, January 1986.

  [4] Kumar, A., Postel, J., Neuman, C., Danzig, P., and S. Miller,
      "Common DNS Implementation Errors and Suggested Fixes", RFC 1536,
      USC/Information Sciences Institute, USC, October 1993.

  [5] Beertema, P., "Common DNS Data File Configuration Errors", RFC
      1537, CWI, October 1993.

Security Considerations

  This memo indicates vulnerabilities with all too-forgiving DNS
  clients.  It points out a correction that would eliminate the future
  potential of the problem.

Author's Address

  Ehud Gavron
  ACES Research Inc.
  PO Box 14546
  Tucson, AZ 85711

  Phone: (602) 743-9841
  EMail: [email protected]

















Gavron                                                          [Page 5]

You are viewing proxied material from gopher.fnord.one. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.