|Summary:||net-wireless/bluez-utils<= 2.19 security vulnerability|
|Product:||Gentoo Security||Reporter:||Henryk Plötz <henryk>|
|Component:||Vulnerabilities||Assignee:||Gentoo Security <security>|
|Package list:||Runtime testing required:||---|
Description Henryk Plötz 2005-08-06 10:10:46 UTC
There is a security vulnerability in bluez-utils 2.16 through 2.18 (I only tested 2.18, but the CVS logs indicate that the problem exists at least since 2.16) which allows for remote command execution (over the air) with root privileges and bypassing of the Bluetooth PIN. The vulnerability occurs when the pin helper is called and thus exists only when "security user" is set in hcid.conf (which is the default in Gentoo Linux, but not in the upstream package). The attacker also needs to get his device name into the device name cache and I'm not sure how to do this in general so it might be harder to exploit in a default installation. However, upstream has released a fixed bluez-utils-2.19 (which also needs bluez-libs-2.19, unfortunately) and I think Gentoo include this version ASAP. Should this not be possible then at least the existing version ought to be patched. The difference between the vulnerable and the non-vulnerable version is: http://cvs.sourceforge.net/viewcvs.py/bluez/utils/hcid/security.c?r1=1.31&r2=1.34 Reproducible: Always Steps to Reproduce:
Comment 1 Thierry Carrez (RETIRED) 2005-08-06 11:45:28 UTC
liquidx or mobile/pda herds: please apply patch or bump to 2.19
Comment 2 Henrik Brix Andersen 2005-08-08 10:19:43 UTC
liquidx: are you around to handle this? If I do not hear from you by tomorrow, I will handle the bump so we can get it marked stable on all affected archs.
Comment 3 Henrik Brix Andersen 2005-08-09 11:59:23 UTC
I have added net-wireless/bluez-libs-2.19 and net-wireless/bluez-utils-2.19, which address the issue of this bug report. I will mark them stable on x86 tomorrow if no additional bug reports tick in. Other archs: please follow.
Comment 4 Thierry Carrez (RETIRED) 2005-08-10 00:55:52 UTC
I asked for a CAN number to MITRE.
Comment 5 Henrik Brix Andersen 2005-08-10 03:49:03 UTC
Stable on x86.
Comment 6 Tobias Scherbaum (RETIRED) 2005-08-10 08:09:24 UTC
Comment 7 Thierry Carrez (RETIRED) 2005-08-12 00:25:02 UTC
====================================================== Candidate: CAN-2005-2547 URL: http://cve.mitre.org/cgi-bin/cvename.cgi?name=CAN-2005-2547 Reference: MLIST:[bluez-devel] 20050804 Possible security vulnerability in hcid when calling pin helper Reference: URL:http://sourceforge.net/mailarchive/forum.php?thread_id=7893206&forum_id=1881 Reference: CONFIRM:https://bugs.gentoo.org/show_bug.cgi?id=101557 Reference: CONFIRM:http://cvs.sourceforge.net/viewcvs.py/bluez/utils/hcid/security.c?r1=1.31&r2=1.34 security.c in hcid for BlueZ 2.18 and earlier allows remote attackers to execute arbitrary commands via shell metacharacters in the Bluetooth device name when invoking the PIN helper.
Comment 8 Carlos Silva (RETIRED) 2005-08-12 03:56:21 UTC
Comment 9 Henrik Brix Andersen 2005-08-12 04:06:17 UTC
What about hppa?
Comment 10 Thierry Carrez (RETIRED) 2005-08-12 08:22:00 UTC
You're right, we missed hppa. hppa, sparc : please test and mark stable.
Comment 11 Michael Hanselmann (hansmi) (RETIRED) 2005-08-13 00:33:13 UTC
Stable on hppa.
Comment 12 Sune Kloppenborg Jeppesen (RETIRED) 2005-08-15 09:47:58 UTC
sparc please test and mark stable ASAP, thx.
Comment 13 Gustavo Zacarias (RETIRED) 2005-08-17 06:14:29 UTC
sparc stable. didn't get the chance to properly test it, but it should be fine, hopefully i'll get the bt kit by this weekend, play with it and leave it be or mask it.
Comment 14 Sune Kloppenborg Jeppesen (RETIRED) 2005-08-17 09:43:15 UTC