Bug 522982 (CVE-2014-3635) – <sys-apps/dbus-1.8.8: Multiple vulnerabilities (CVE-2014-{3635,3636,3637,3638,3639)

• Product: Gentoo Security
• Status: RESOLVED FIXED

[Description Samuli Suominen (RETIRED) 2014-09-17 06:07:22 UTC]

+The "smashy smashy egg man" release.
+
+Security fixes:
+
+• Do not accept an extra fd in the padding of a cmsg message, which
+ could lead to a 4-byte heap buffer overrun.
+ (CVE-2014-3635, fd.o #83622; Simon McVittie)
+
+• Reduce default for maximum Unix file descriptors passed per message
+ from 1024 to 16, preventing a uid with the default maximum number of
+ connections from exhausting the system bus' file descriptors under
+ Linux's default rlimit. Distributors or system administrators with a
+ more restrictive fd limit may wish to reduce these limits further.
+
+ Additionally, on Linux this prevents a second denial of service
+ in which the dbus-daemon can be made to exceed the maximum number
+ of fds per sendmsg() and disconnect the process that would have
+ received them.
+ (CVE-2014-3636, fd.o #82820; Alban Crequy)
+
+• Disconnect connections that still have a fd pending unmarshalling after
+ a new configurable limit, pending_fd_timeout (defaulting to 150 seconds),
+ removing the possibility of creating an abusive connection that cannot be
+ disconnected by setting up a circular reference to a connection's
+ file descriptor.
+ (CVE-2014-3637, fd.o #80559; Alban Crequy)
+
+• Reduce default for maximum pending replies per connection from 8192 to 128,
+ mitigating an algorithmic complexity denial-of-service attack
+ (CVE-2014-3638, fd.o #81053; Alban Crequy)
+
+• Reduce default for authentication timeout on the system bus from
+ 30 seconds to 5 seconds, avoiding denial of service by using up
+ all unauthenticated connection slots; and when all unauthenticated
+ connection slots are used up, make new connection attempts block
+ instead of disconnecting them.
+ (CVE-2014-3639, fd.o #80919; Alban Crequy)

[Comment 1 Samuli Suominen (RETIRED) 2014-09-17 06:10:38 UTC]

Please test and stabilize:

=sys-apps/dbus-1.8.8 alpha amd64 arm hppa ia64 ppc ppc64 sparc x86

[Comment 2 Jeroen Roovers (RETIRED) 2014-09-17 09:50:17 UTC]

Stable for HPPA.

[Comment 3 Agostino Sarubbo 2014-09-17 12:08:35 UTC]

amd64 stable

[Comment 4 Agostino Sarubbo 2014-09-17 12:08:49 UTC]

x86 stable

[Comment 5 Agostino Sarubbo 2014-09-19 10:35:08 UTC]

sparc stable

[Comment 6 Agostino Sarubbo 2014-09-19 10:36:51 UTC]

ppc64 stable

[Comment 7 Agostino Sarubbo 2014-09-20 11:05:51 UTC]

alpha stable

[Comment 8 Markus Meier 2014-09-21 20:18:18 UTC]

arm stable

[Comment 9 Agostino Sarubbo 2014-09-27 10:39:10 UTC]

ia64 stable

[Comment 10 GLSAMaker/CVETool Bot 2014-10-05 13:13:37 UTC]

CVE-2014-3639 (http://nvd.nist.gov/nvd.cfm?cvename=CVE-2014-3639):
  The dbus-daemon in D-Bus before 1.6.24 and 1.8.x before 1.8.8 does not
  properly close old connections, which allows local users to cause a denial
  of service (incomplete connection consumption and prevention of new
  connections) via a large number of incomplete connections.

CVE-2014-3638 (http://nvd.nist.gov/nvd.cfm?cvename=CVE-2014-3638):
  The bus_connections_check_reply function in config-parser.c in D-Bus before
  1.6.24 and 1.8.x before 1.8.8 allows local users to cause a denial of
  service (CPU consumption) via a large number of method calls.

CVE-2014-3637 (http://nvd.nist.gov/nvd.cfm?cvename=CVE-2014-3637):
  D-Bus 1.3.0 through 1.6.x before 1.6.24 and 1.8.x before 1.8.8 does not
  properly close connections for processes that have terminated, which allows
  local users to cause a denial of service via a D-bus message containing a
  D-Bus connection file descriptor.

CVE-2014-3635 (http://nvd.nist.gov/nvd.cfm?cvename=CVE-2014-3635):
  Off-by-one error in D-Bus 1.3.0 through 1.6.x before 1.6.24 and 1.8.x before
  1.8.8, when running on a 64-bit system and the max_message_unix_fds limit is
  set to an odd number, allows remote attackers to cause a denial of service
  (dbus-daemon crash) or possibly execute arbitrary code by sending one more
  file descriptor than the limit, which triggers a heap-based buffer overflow
  or an assertion failure.

[Comment 11 Agostino Sarubbo 2014-10-05 15:07:17 UTC]

ppc stable.

Maintainer(s), please cleanup.
Security, please add it to the existing request, or file a new one.

[Comment 12 Mikle Kolyada (RETIRED) 2014-12-13 14:21:23 UTC]

Added to existing glsa draft.

[Comment 13 GLSAMaker/CVETool Bot 2014-12-13 15:14:43 UTC]

This issue was resolved and addressed in
 GLSA 201412-12 at http://security.gentoo.org/glsa/glsa-201412-12.xml
by GLSA coordinator Mikle Kolyada (Zlogene).