The processchpwrequest function in schpw.c in the password-changing functionality in kadmind in MIT Kerberos 5 (aka krb5) 1.7 through 1.9 frees an invalid pointer, which allows remote attackers to execute arbitrary code or cause a denial of service (daemon crash) via a crafted request that triggers an error condition.

Integer underflow flaws, leading to heap-based corruption, were found in the way MIT Kerberos Key Distribution Center (KDC) used to decrypt ciphertexts encrypted with Advanced Encryption Standard (AES) and ARCFOUR (RC4) encryption algorithms. A remote KDC client could provide a specially-crafted AES or RC4 encrypted ciphertext(s), which once processed by a central KDC would lead to denial of service (KDC crash or abort) or, potentially, to KDC relying authentication services compromise or arbitrary code execution with the privileges of the KDC centre.

Upstream advisory (not available yet): -------------------------------------- http://web.mit.edu/kerberos/advisories/MITKRB5-SA-2009-004.txt

Upstream patches (not available yet): ------------------------------------- http://web.mit.edu/kerberos/advisories/2009-004-patch1.7.txt http://web.mit.edu/kerberos/advisories/2009-004-patch1.6.3.txt

Buffer overflow in the RPC library used by libgssrpc and kadmind in MIT Kerberos 5 (krb5) 1.4 through 1.6.3 allows remote attackers to execute arbitrary code by triggering a large number of open file descriptors.

Double free vulnerability in MIT Kerberos 5 (aka krb5) allows attackers to have unspecified impact via vectors involving automatic deletion of security contexts on error.

KDC in MIT Kerberos 5 (krb5kdc) does not set a global variable for some krb4 message types, which allows remote attackers to cause a denial of service (crash) and possibly execute arbitrary code via crafted messages that trigger a NULL pointer dereference or double-free.

Buffer overflow in the RPC library (lib/rpc/rpcdtablesize.c) used by libgssrpc and kadmind in MIT Kerberos 5 (krb5) 1.2.2, and probably other versions before 1.3, when running on systems whose unistd.h does not define the FDSETSIZE macro, allows remote attackers to cause a denial of service (crash) and possibly execute arbitrary code by triggering a large number of open file descriptors.

The krb5gssprocesscontexttoken function in lib/gssapi/krb5/processcontexttoken.c in the libgssapikrb5 library in MIT Kerberos 5 (aka krb5) through 1.11.5, 1.12.x through 1.12.2, and 1.13.x before 1.13.1 does not properly maintain security-context handles, which allows remote authenticated users to cause a denial of service (use-after-free and double free, and daemon crash) or possibly execute arbitrary code via crafted GSSAPI traffic, as demonstrated by traffic to kadmind.

PAC parsing in MIT Kerberos 5 (aka krb5) before 1.19.4 and 1.20.x before 1.20.1 has integer overflows that may lead to remote code execution (in KDC, kadmind, or a GSS or Kerberos application server) on 32-bit platforms (which have a resultant heap-based buffer overflow), and cause a denial of service on other platforms. This occurs in krb5pacparse in lib/krb5/krb/pac.c. Heimdal before 7.7.1 has "a similar bug."

The iakerbgssexportseccontext function in lib/gssapi/krb5/iakerb.c in MIT Kerberos 5 (aka krb5) 1.14 pre-release 2015-09-14 improperly accesses a certain pointer, which allows remote authenticated users to cause a denial of service (memory corruption) or possibly have unspecified other impact by interacting with an application that calls the gssexportseccontext function. NOTE: this vulnerability exists because of an incorrect fix for CVE-2015-2696.

The krb5ldaplockoutaudit function in the Key Distribution Center (KDC) in MIT Kerberos 5 (aka krb5) 1.8 through 1.8.4 and 1.9 through 1.9.1, when the LDAP back end is used, allows remote attackers to cause a denial of service (assertion failure and daemon exit) via unspecified vectors, related to the lockedcheckp function. NOTE: the Berkeley DB vector is covered by CVE-2011-4151.

The kdbldap plugin in the Key Distribution Center (KDC) in MIT Kerberos 5 (aka krb5) 1.9 through 1.9.1, when the LDAP back end is used, allows remote attackers to cause a denial of service (NULL pointer dereference and daemon crash) via a kinit operation with incorrect string case for the realm, related to the isprincipalinrealm, krb5seterrormessage, krb5ldapgetprincipal, and processasreq functions.

The lookuplockoutpolicy function in the Key Distribution Center (KDC) in MIT Kerberos 5 (aka krb5) 1.8 through 1.8.4 and 1.9 through 1.9.1, when the db2 (aka Berkeley DB) or LDAP back end is used, allows remote attackers to cause a denial of service (NULL pointer dereference and daemon crash) via vectors that trigger certain processasreq errors.

The krb5db2lockoutaudit function in the Key Distribution Center (KDC) in MIT Kerberos 5 (aka krb5) 1.8 through 1.8.4, when the db2 (aka Berkeley DB) back end is used, allows remote attackers to cause a denial of service (assertion failure and daemon exit) via unspecified vectors, a different vulnerability than CVE-2011-1528.

Double free vulnerability in the prepareerroras function in doasreq.c in the Key Distribution Center (KDC) in MIT Kerberos 5 (aka krb5) 1.7 through 1.9, when the PKINIT feature is enabled, allows remote attackers to cause a denial of service (daemon crash) or possibly execute arbitrary code via an edata field containing typed data.

The (1) kadm5createprincipal3 and (2) kadm5modifyprincipal functions in lib/kadm5/srv/svrprincipal.c in kadmind in MIT Kerberos 5 (aka krb5) 1.12.x and 1.13.x before 1.13.4 and 1.14.x before 1.14.1 allow remote authenticated users to cause a denial of service (NULL pointer dereference and daemon crash) by specifying KADM5POLICY with a NULL policy name.

ecverify in kdc/kdcpreauthec.c in the Key Distribution Center (KDC) in MIT Kerberos 5 (aka krb5) before 1.18.4 and 1.19.x before 1.19.2 allows remote attackers to cause a NULL pointer dereference and daemon crash. This occurs because a return value is not properly managed in a certain situation.

The Kerberos 4 support in KDC in MIT Kerberos 5 (krb5kdc) does not properly clear the unused portion of a buffer when generating an error message, which might allow remote attackers to obtain sensitive information, aka "Uninitialized stack values."

A flaw was found in, Fedora versions of krb5 from 1.16.1 to, including 1.17.x, in the way a Kerberos client could crash the KDC by sending one of the RFC 4556 "enctypes". A remote unauthenticated user could use this flaw to crash the KDC.

lib/gssapi/krb5/iakerb.c in MIT Kerberos 5 (aka krb5) before 1.14 relies on an inappropriate context handle, which allows remote attackers to cause a denial of service (incorrect pointer read and process crash) via a crafted IAKERB packet that is mishandled during a gssinquirecontext call.

A denial of service flaw was found in Kerberos's GSS-API spnego security mechanism implementation. A remote attacker could use this flaw to cause gss-server crash via invalid ContextFlags for the reqFlags field in the NegTokenInit in spnegomech.c, which triggers an assertion failure. Similar vulnerability than CVE-2009-0845.

PGP-signed patch from upstream will be available at: http://web.mit.edu/kerberos/advisories/2010-002-patch.txt.asc

A null pointer dereference flaw was found in Kerberos's GSS-API spnego security mechanism implemenation. A local user could use this flaw to cause a denial of service (krb5 daemon crash) via invalid ContextFlags for the reqFlags field in the NegTokenInit (RFC 4178).

References: http://krbdev.mit.edu/rt/Ticket/Display.html?user=guest&pass=guest&id=6402

Upstream patch: http://src.mit.edu/fisheye/changelog/krb5/?cs=22099

It was found that the kerberized FTP server did not properly check for the failure to set its effective group identifier (GID). A remote, authenticated FTP user could use this flaw to gain unauthorized read or write access to files whose group owner was the initial effective GID of the FTP daemon process.

References: [1] http://web.mit.edu/kerberos/advisories/MITKRB5-SA-2011-005.txt (not public yet) [2] http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2011-1526

Upstream patch: [3] http://web.mit.edu/kerberos/advisories/2011-005-patch.txt

Acknowledgements:

Red Hat would like to thank the MIT Kerberos project for reporting this issue. Upstream acknowledges Tim Zingelman as the original reporter.

Multiple memory leaks in kadmin/server/serverstubs.c in kadmind in MIT Kerberos 5 (aka krb5) before 1.13.4 and 1.14.x before 1.14.1 allow remote authenticated users to cause a denial of service (memory consumption) via a request specifying a NULL principal name.

The mergeauthdata function in kdcauthdata.c in the Key Distribution Center (KDC) in MIT Kerberos 5 (aka krb5) 1.8.x before 1.8.4 does not properly manage an index into an authorization-data list, which allows remote attackers to cause a denial of service (daemon crash), or possibly obtain sensitive information, spoof authorization, or execute arbitrary code, via a TGS request that triggers an uninitialized pointer dereference, as demonstrated by a request from a Windows Active Directory client.

The validateasrequest function in kdcutil.c in the Key Distribution Center (KDC) in MIT Kerberos 5 (aka krb5) before 1.13.6 and 1.4.x before 1.14.3, when restrictanonymoustotgt is enabled, uses an incorrect client data structure, which allows remote authenticated users to cause a denial of service (NULL pointer dereference and daemon crash) via an S4U2Self request.

lib/kadm5/kadmrpcxdr.c in MIT Kerberos 5 (aka krb5) before 1.20.2 and 1.21.x before 1.21.1 frees an uninitialized pointer. A remote authenticated user can trigger a kadmind crash. This occurs because xdrkadm5principalentrec does not validate the relationship between nkeydata and the keydata array count.

MIT Kerberos 5 (aka krb5) 1.8.x through 1.8.3 does not reject RC4 key-derivation checksums, which might allow remote authenticated users to forge a (1) AD-SIGNEDPATH or (2) AD-KDC-ISSUED signature, and possibly gain privileges, by leveraging the small key space that results from certain one-byte stream-cipher operations.

The kdcpreauth modules in MIT Kerberos 5 (aka krb5) 1.12.x and 1.13.x before 1.13.2 do not properly track whether a client's request has been validated, which allows remote attackers to bypass an intended preauthentication requirement by providing (1) zero bytes of data or (2) an arbitrary realm name, related to plugins/preauth/otp/main.c and plugins/preauth/pkinit/pkinitsrv.c.

Kerberos 5 (aka krb5) 1.21.2 contains a memory leak vulnerability in /krb5/src/kdc/ndr.c.

It was reported [1] that, within the kadmin protocol, the access controls for getstrings/setstring were insufficient; anyone with global list privileges could get or modify string attributed on any principal.

It was also noted that the exposure depends on how generous the kadmind acl was with list permissions and whether or not string attributes were used in deployment (and noting that nothing in the core code uses them yet).

This has been fixed upstream [2] and in Fedora [3].

[1] http://krbdev.mit.edu/rt/Ticket/Display.html?user=guest&pass=guest&id=7093 [2] http://src.mit.edu/fisheye/changelog/krb5/?cs=25704 [3] http://koji.fedoraproject.org/koji/buildinfo?buildID=300840