==Phrack Inc.==
Volume 0x0b, Issue 0x3e, Phile #0x03 of 0x10
|=-----------------------------------------------------------------------=|
|=---------------------=[ L I N E N O I S E ]=---------------------------=|
|=-----------------------------------------------------------------------=|
1 - Mistakes in the RFC Guidelines on DNS Spoofing Attacks
2 - Injecting Signals by Shaun
3 - Pirating A Radio Station
|=------=[ The Impact of RFC Guidelines on DNS Spoofing Attacks ]=------=|
by have2Banonymous
--[ Contents
1 - Executive Summary
2 - Overview of Basic DNS Spoofing Attacks
3 - Proposed Criteria for DNS Reply Acceptance
4 - Impact of RFC Guidelines on DNS Reply Acceptance Criteria
5 - Example DNS Spoofing Attack
6 - Practical Impact of RFC Guidelines on DNS Spoofing Attacks
7 - Implementation Comparison
8 - Conclusion
--[ 1 - Executive Summary
This article provides a brief overview of basic Domain Name System (DNS)
spoofing attacks against DNS client resolvers. Technical challenges are
proposed that should help to both identify attempted attacks and prevent
them from being successful. Relevant Request for Comments (RFC)
guidelines, used by programmers to help ensure their DNS resolver code
meets specifications, are reviewed. This results in the realisation that
the RFC guidelines are not adequately specific or forceful to help
identify or prevent DNS spoofing attacks against DNS client resolvers.
Furthermore, the RFC guidelines actually simplify such attacks to a level
that has not previously been discussed in the public domain until now.
To highlight the consequences of merely conforming to the RFC guidelines
without considering security ramifications, an example DNS spoofing attack
against the DNS resolver in Microsoft Windows XP is provided. This
illustrates serious weaknesses in the Windows XP DNS resolver client
implementation. For example, Windows XP will accept a DNS reply as being
valid without performing a thorough check that the DNS reply actually
matches the DNS request. This allows an attacker to create malicious
generic DNS replies that only need to meet a couple of criteria with
predictable values in order to be accepted as a valid DNS reply by the
targeted user.
This article discusses the practical impact of the issues raised, such as
the ability to perform a successful and reasonably undetectable DNS
spoofing attack against a large target base of Windows XP users, without
the attacker requiring knowledge of the DNS requests issued by the
targeted users. Finally, a comparison with the DNS resolver in Debian
Linux is supplied.
--[ 2 - Overview of Basic DNS Spoofing Attacks
When a user types the web site name www.somewebsite.org into their web
browser, their computer issues a DNS request to their Internet Service
Provider's (ISP) DNS server to resolve the web site name to an IP address.
An attacker may attempt to subvert this process by sending the user a DNS
reply containing an incorrect IP address, resulting in the user's computer
connecting to a computer of the attacker's choice instead of the desired
web site.
--[ 3 - Proposed Criteria for DNS Reply Acceptance
RFC 2535 (Domain Name System Security Extensions) otherwise known as
DNSSEC discusses how cryptographic digital signatures can be used to
authenticate DNS transactions to help mitigate DNS spoofing attacks.
However, the adoption of this technology has been extremely slow. Even
without this level of security, it would initially appear that a DNS
spoofing attack against a DNS client resolver would be challenging to
perform. This challenge results from the following proposed criteria of
the DNS reply that must be met for it to be accepted by the computer
performing the DNS lookup.
Proposed criteria of a DNS reply for it to be accepted:
1) The source IP address must match the IP address that the DNS request
was sent to.
2) The destination IP address must match the IP address that the DNS
request was sent from.
3) The source port number must match the port number that the DNS request
was sent to.
4) The destination port number must match the port number that the DNS
request was sent from.
5) The UDP checksum must be correctly calculated. This may require the
attacker to spend more time and effort per attack, although some packet
generation utilities have the ability to automatically calculate this
value.
6) The transaction ID must match the transaction ID in the DNS request.
7) The domain name in the question section must match the domain name in
the question section of the DNS request.
8) The domain name in the answer section must match the domain name in the
question section of the DNS request.
9) The requesting computer must receive the attacker's DNS reply before it
receives the legitimate DNS reply.
--[ 4 - Impact of RFC Guidelines on DNS Reply Acceptance Criteria
According to the RFC guidelines, it is not necessary for all of these
criteria to be met in order for a DNS reply to be accepted. Specifically,
criteria 1, 2, 3, 5, 7 and 8 do not have to be met, while criteria 4, 6
and 9 must be met. The following is a devil's advocate interpretation of
the RFC guidelines and a detailed discussion of their effect on each
criteria.
Criteria 1 (source IP address) does not have to be met according to RFC
791 (Internet Protocol) which states that "In general, an implementation
must be conservative in its sending behavior, and liberal in its receiving
behavior. That is, it must be careful to send well-formed datagrams, but
must accept any datagram that it can interpret (e.g., not object to
technical errors where the meaning is still clear)". RFC 1035 (Domain
names - implementation and specification) states that "Some name servers
send their responses from different addresses than the one used to receive
the query. That is, a resolver cannot rely that a response will come from
the same address which it sent the corresponding query to". The source IP
address can therefore be set to an arbitrary IP address. Regardless, if
desired, the attacker can set the source IP address of their DNS replies
to that of the targeted user's DNS server. This is especially easy if the
targeted user is a dialup ISP user since the ISP may have a friendly "How
to setup your Internet connection" web page that specifies the IP address
of their DNS server.
Criteria 2 (destination IP address) does not have to be met according to
RFC 1122 (Requirements for Internet Hosts -- Communication Layers) which
states that "For most purposes, a datagram addressed to a broadcast or
multicast destination is processed as if it had been addressed to one of
the host's IP addresses". Using a broadcast destination address would be
most useful for attacking computers on a Local Area Network. Furthermore,
a DNS reply may be accepted if it is addressed to any of the IP addresses
associated with a network interface.
Criteria 3 (source port number) does not have to be met according to RFC
768 (User Datagram Protocol) which states that "Source Port is an optional
field". The source port can therefore be set to an arbitrary value such
as 0 or 12345. Since the source port number of the DNS reply affects
packet dissection by utilities such as Ethereal, a value of 137 is a
devious choice since it will be dissected as the NetBIOS Name Service
(NBNS) protocol which is based on DNS. As a result, the malicious DNS
replies can be made to appear like NetBIOS traffic which is likely to be
discarded by the system administrator or investigator as typical NetBIOS
background noise.
Criteria 4 (destination port number) must be met according to RFC 768
(User Datagram Protocol). However, this value may be predictable
depending on the requesting computer's operating system. During testing,
Windows XP always used port number 1026 to perform DNS queries, though
this value depends on when the DNS Client service started during the boot
process.
Criteria 5 (UDP checksum) does not have to be met according to RFC 1122
(Requirements for Internet Hosts -- Communication Layers) which states
that "the UDP checksum is optional; the value zero is transmitted in the
checksum field of a UDP header to indicate the absence of a checksum".
Criteria 6 (transaction ID) must be met according to RFC 1035 (Domain
names - implementation and specification) which states that the
transaction ID is used "to match up replies to outstanding queries".
However, this value may be predictable depending on the requesting
computer's operating system. During testing, Windows XP did not randomly
choose the 16 bit transaction ID value. Rather, Windows XP always used a
transaction ID of 1 for the first DNS query performed after the computer
was turned on, with the transaction ID simply incremented for subsequent
DNS queries. Transaction ID 1 and 2 were used by the operating system to
perform a DNS query of time.windows.com.
Criteria 7 and 8 (domain name in question and answer section) do not have
to be met according to RFC 1035 (Domain names - implementation and
specification) which states that the transaction ID is used "to match up
replies to outstanding queries" and recommends as a secondary step "to
verify that the question section corresponds to the information currently
desired". RFC recommendations do not have to be followed, and in the case
of an absent question section, the principal that an implementation must
accept any datagram that it can interpret appears to apply. Therefore, a
DNS reply containing a single answer in the form of an IP address can be
matched to the corresponding DNS request based on the transaction ID,
without requiring a question section and without resorting to the overhead
of processing the domain information in the answer section. Furthermore,
an answer section is not even necessary if an Authority section is
provided to refer the requesting computer to an authoritative name server
(or a DNS server under the attacker's control).
Criteria 9 (requesting computer must receive the attacker's DNS reply
before it receives the legitimate DNS reply) must be met and remains as
the greatest challenge to the attacker. This restriction is difficult
to bypass unless the legitimate DNS server is taken out of action to
prevent competition with the spoofed DNS reply, or numerous spoofed DNS
replies are sent to the targeted user. However, as discussed above,
criteria 1 to 8 either do not have to be met or may have predictable
values. Therefore an attacker may require no knowledge of the victim's
DNS request to have a reasonable chance of performing a successful attack
by sending the requesting computer a small number of generic DNS replies.
Furthermore, there is a viable workaround to the restrictive nature of
this criteria. If the attacker is not trying to compromise a specific
computer, a "spray and pray" approach can be used. This approach involves
sending a very small number (twenty) of spoofed DNS replies to a maximum
number of potential target computers, instead of trying to compromise a
specific user and only once they have been compromised then trying to
compromise another specific user. This "spray and pray" approach won't
compromise every potential victim, and every packet the attacker sends
won't result in a compromise, but enough of the attacker's malicious DNS
replies will be accepted by enough potential victims to make the exercise
worthwhile.
--[ 5 - Example DNS Spoofing Attack
A DNS spoofing attack using the concepts discussed in this article was
performed against a Windows XP computer. The test Windows XP computer
was a default install of the operating system followed by the application
of Service Pack 1. The Microsoft Internet Connection Firewall shipped
with Windows XP was then enabled, and configured to perform full logging
of dropped packets and successful connections.
The Windows XP user typed the web site URL www.somewebsite.org into
Internet Explorer, resulting in a DNS request being sent from the user's
computer (IP address 192.168.1.1) to the user's DNS server (IP address
192.168.1.254).
A spoofed DNS reply disguised as NetBIOS data was sent to the user from
the fake (spoofed) nonexistent IP address 10.10.10.1, specifying that
whatever name the user was attempting to resolve had the IP address
192.168.1.77. The IP address 192.168.1.77 was actually a web server
under the attacker's control.
Internet Explorer connected to 192.168.1.77 and requested the web page.
This revealed that the designers of the DNS resolver in Microsoft Windows
XP also interpreted the RFC guidelines as described in the previous
section, significantly simplifying DNS spoofing attacks.
The following network packet decoded by Ethereal version 0.10.3
illustrates the malicious DNS reply and demonstrates how Ethereal can be
confused into decoding the packet as NetBIOS traffic.
Frame 1 (102 bytes on wire, 102 bytes captured)
Ethernet II, Src: 00:50:56:c0:00:01, Dst: 00:0c:29:04:7d:25
Internet Protocol, Src Addr: 10.10.10.1 (10.10.10.1), Dst Addr:
192.168.1.1 (192.168.1.1)
User Datagram Protocol, Src Port: 137 (137), Dst Port: 1026 (1026)
Source port: 137 (137)
Destination port: 1026 (1026)
Length: 68
Checksum: 0x0000 (none)
NetBIOS Name Service
Transaction ID: 0x0003
Flags: 0x8580 (Name query response, No error)
Questions: 0
Answer RRs: 1
Authority RRs: 0
Additional RRs: 0
Answers
WORKGROUP<1b>: type unknown, class inet
Name: WORKGROUP<1b>
Type: unknown
Class: inet
Time to live: 1 day
Data length: 4
Data
0000 00 0c 29 04 7d 25 00 50 56 c0 00 01 08 00 45 00 ..).}%.PV.....E.
0010 00 58 bf 58 00 00 00 11 25 89 0a 0a 0a 01 c0 a8 .X.X....%.......
0020 01 01 00 89 04 02 00 44 00 00 00 03 85 80 00 00 .......D........
0030 00 01 00 00 00 00 20 46 48 45 50 46 43 45 4c 45 ...... FHEPFCELE
0040 48 46 43 45 50 46 46 46 41 43 41 43 41 43 41 43 HFCEPFFFACACACAC
0050 41 43 41 43 41 42 4c 00 00 01 00 01 00 01 51 80 ACACABL.......Q.
0060 00 04 c0 a8 01 4d .....M
This packet was created using the following parameters passed to the
freely available netwox packet creation utility:
netwox 38 --ip4-src 10.10.10.1 --ip4-dst 192.168.1.1 --ip4-protocol 17
--ip4-data 008904020044000000038580000000010000000020464845504643454c45484
643455046464641434143414341434143414341424c0000010001000151800004c0a8014d
Alternatively, the following parameters could be used since netwox
automatically calculates the UDP checksum:
netwox 39 --ip4-src 10.10.10.1 --ip4-dst 192.168.1.1 --udp-src 137
--udp-dst 1026 --udp-data 00038580000000010000000020464845504643454c45484
643455046464641434143414341434143414341424c0000010001000151800004c0a8014d
The following shows that the spoofed DNS reply has been added to the
user's DNS resolver cache for a period of 1 day, causing future
resolutions of www.somewebsite.org to map to the web server under the
attacker's control. The cache duration value can be decreased by the
attacker so that the entry is either not cached or is immediately removed
from the cache in order to remove evidence of the attack.
C:\>ipconfig /displaydns
Windows IP Configuration
1.0.0.127.in-addr.arpa
----------------------------------------
Record Name . . . . . : 1.0.0.127.in-addr.arpa.
Record Type . . . . . : 12
Time To Live . . . . : 604393
Data Length . . . . . : 4
Section . . . . . . . : Answer
PTR Record . . . . . : localhost
www.somewebsite.org
----------------------------------------
Record Name . . . . . : FHEPFCELEHFCEPFFFACACACACACACABL
Record Type . . . . . : 1
Time To Live . . . . : 86364
Data Length . . . . . : 4
Section . . . . . . . : Answer
A (Host) Record . . . : 192.168.1.77
localhost
----------------------------------------
Record Name . . . . . : localhost
Record Type . . . . . : 1
Time To Live . . . . : 604393
Data Length . . . . . : 4
Section . . . . . . . : Answer
A (Host) Record . . . : 127.0.0.1
The following log file from Microsoft's Internet Connection Firewall
reveals that it did not provide any protection against the attack, though
it is not designed to inspect and correlate DNS traffic. If the firewall
was not configured to log successful connections, then there would not
have been any log entries.
#Verson: 1.0
#Software: Microsoft Internet Connection Firewall
#Time Format: Local
#Fields: date time action protocol src-ip dst-ip src-port dst-port size
tcpflags tcpsyn tcpack tcpwin icmptype icmpcode info
2004-05-10 20:34:56 OPEN UDP 192.168.1.1 192.168.1.254 1026 53 - - - - - -
- -
2004-05-10 20:34:57 OPEN-INBOUND UDP 10.10.10.1 192.168.1.1 137 1026 - - -
- - - - -
2004-05-10 20:34:57 OPEN TCP 192.168.1.1 192.168.1.77 3010 80 - - - - - -
- -
2004-05-10 20:35:30 CLOSE TCP 192.168.1.1 192.168.1.77 3010 80 - - - - - -
- -
2004-05-10 20:36:30 CLOSE UDP 192.168.1.1 192.168.1.254 1026 53 - - - - -
- - -
2004-05-10 20:36:30 CLOSE UDP 10.10.10.1 192.168.1.1 137 1026 - - - - - -
- -
It can be seen that when the Windows XP computer sent a UDP packet from
port 1026 to port 53 of the DNS server, the firewall allowed all incoming
UDP traffic to port 1026, regardless of the source IP address or source
port of the incoming traffic. Such incoming traffic was allowed to
continue until the firewall decided to block access to port 1026, which
occurred when there was no incoming traffic to port 1026 for a defined
period of time. This timeframe was between 61 seconds and 120 seconds, as
it appeared that the firewall checked once per minute to determine if
access to ports should be revoked due to more than 60 seconds of
inactivity. Assuming that users connected to the Internet would typically
perform a DNS query at least every minute, incoming access to port 1026
would always be granted. An attacker on the Internet could therefore send
the Windows XP computer spoofed DNS replies without worrying that they
might be blocked by the firewall. Such traffic would not generate any
logs if the firewall was configured to only Log Dropped Packets. If the
firewall was configured to also Log Successful Connections as in this
example, these log entries would disappear among the thousands of other
log entries. Since the firewall logs connections and not traffic, if the
source IP address was set to the Windows XP computer's DNS server, no
extra firewall log entries would be created as a result of the DNS
spoofing attack.
The netstat command revealed that the Windows XP computer was always
listening on UDP port 1026, and as a result, extra DNS replies were
silently discarded and did not generate an error message in the event log
or an ICMP port unreachable packet. This behaviour, and the reuse of the
same source port number for DNS requests, was attributed to the DNS Client
service.
--[ 6 - Practical Impact of RFC Guidelines on DNS Spoofing Attacks
The attacker does not require information about the targeted user's DNS
requests, such as the IP address of the user's DNS server, the source port
of the user's DNS request, or the name that the user was attempting to
resolve to an IP address. Therefore the attacker does not require access
to the communication link between the targeted user and their DNS server.
Windows XP SP1 matches DNS replies to DNS requests by only the transaction
ID and the UDP port number, and both of these values are very predictable.
Since the name to be resolved is not matched between the DNS request and
the DNS reply, the attacker does not care what domain name the user
queried since this domain name does not have to be placed in the
attacker's DNS reply. As a result, the attacker can create generic
malicious DNS replies that will successfully subvert the targeted user's
DNS lookup process regardless of the name the targeted user was attempting
to resolve, and regardless of the targeted user's network configuration
such as the IP address of their DNS server.
An attacker desiring to compromise as many computers as possible with the
least amount of effort and in the shortest timeframe could send twenty DNS
replies that look similar to the generic DNS reply used in the example
attack on Windows XP in this article, though with the transaction ID
ranging from 3 to 22. To be more thorough, the attacker could instead
send one hundred DNS replies with the destination port number ranging from
1025 to 1029. The attacker would use a "spray and pray" approach by
sending these DNS replies to every IP address in the IP address range
belonging to a large dialup Internet Service Provider, and when finished,
repeating the process.
A level of success is guaranteed in such an attack scenario considering
the huge target base of potential victims awaiting a DNS reply, and
considering that Windows XP accepts anything vaguely resembling a DNS
reply as a valid DNS reply.
A recipient of the attacker's twenty DNS replies will accept one of them
as being valid, resulting in a successful attack, if the recipient:
- is using Windows XP with its poorly implemented DNS client resolver
(most dialup Internet users are in this category).
- recently connected to the Internet within the last 10-20 minutes or so
and therefore haven't performed more than twenty DNS requests (a
reasonable proportion of dialup Internet users are in this category).
- recently performed a DNS request and is awaiting a DNS reply (a
reasonable number of the huge target base of dialup Internet users are
in this category).
The targeted Windows XP users would be unlikely to notice the attack,
especially if they were relying on Microsoft Internet Connection Firewall
to protect them. Analysis of the logs of a more sophisticated firewall
and inspection of network traffic would not readily reveal a DNS spoofing
attack since the source IP address would not be that of the legitimate DNS
server. Furthermore, the source port number and content of the spoofed
DNS replies can be crafted to make them appear to be typical NetBIOS
background noise which would probably be discarded by the user as useless
network traffic floating around the Internet. Finally, the targeted IP
address range of a dialup ISP would consist mainly of home Internet users
who are not educated in advanced network security concepts.
The IP address in the spoofed DNS replies could be a computer on the
Internet under the attacker's control, which is running proxy software for
email (SMTP and POP3) and HTTP traffic. The attacker would be able to
collect sensitive information including email sent and received as well as
passwords for future email retrieval. Web based email and unencrypted
login details to web sites would also be collected. The attacker could
add content to HTML pages before returning them to the user. Such content
could include banner ads to generate money, or a hidden frame with a link
to a file on a third party web site effectively causing a distributed
denial of service attack against the third party. More seriously, the
attacker could increase the scope of the compromise by adding HTML content
that exploited one of the publicly known vulnerabilities in Internet
Explorer that allows the execution of arbitrary code, but for which there
is no vendor patch. For example, vulnerabilities discussed at the web
site http://www.computerworld.com.au/index.php?id=117316298&eid=-255
The "spray and pray" attack approach is useful for creating a network of
semi-randomly chosen compromised computers under the attacker's control,
otherwise known as a botnet.
Proxying of HTTP/1.1 traffic could be performed by inspecting the HOST
header to determine which web site the user wanted to visit. However, for
the purpose of easily and seamlessly proxying traffic, an attacker may
decide not to place an Answer section in the spoofed DNS replies. Rather,
the attacker may send a non-authoritative spoofed DNS reply using the
Authority and Additional sections of DNS replies to refer the requesting
computer to a DNS server under the attacker's control. This would allow
the attacker to know exactly what domain the victim computer was
attempting to query, and furthermore such spoofed DNS replies may have a
long lasting and widespread effect on the victim's computer. A detailed
discussion of DNS referrals and testing whether Windows XP could handle
them is outside the scope of this article.
--[ 7 - Implementation Comparison
Contributors to the Linux operating system appear to have taken a hardline
security conscious approach to interpreting the RFC guidelines, bordering
on non-conformance for the sake of security. The Mozilla web browser
running on the author's Debian Linux computer was very restrictive and
required DNS replies to meet all of the above nine criteria except for
criteria 5, where a UDP checksum value of zero was accepted. An incorrect
UDP checksum was accepted when the packet was sent over a local network
but not when sent over the Internet. Reviewing the kernel source code
indicated that for local networks, the UDP checksum was deliberately
ignored and hardware based checking was performed instead for performance
reasons. This appeared to be a feature and not a bug, even though it did
not comply with RFC 1122 (Requirements for Internet Hosts -- Communication
Layers) which states that "If a UDP datagram is received with a checksum
that is non-zero and invalid, UDP MUST silently discard the datagram".
During testing, the Linux computer used source port numbers 32768 and
32769 to perform DNS queries. The transaction ID was randomly generated,
complicating DNS spoofing attacks, though the transaction ID used in the
retransmission of an unanswered DNS request was not as random. The choice
of transaction ID values appeared robust enough to help defend against DNS
spoofing attacks on the Internet since the initial transaction ID value
was unpredictable, and the first DNS request would typically be answered
resulting in no need for retransmissions.
The iptables firewall on the Linux computer was configured so that the
only allowed UDP traffic was to/from port 53 of the legitimate DNS server.
When a DNS query was performed and a DNS reply was received, iptables was
unable to block extra (spoofed) incoming DNS replies since it is not
designed to inspect DNS traffic and allow one incoming DNS reply per
outgoing DNS request. However, since the port used to send the DNS query
was closed once a valid DNS reply was received, ICMP port unreachable
messages were generated for the extra (spoofed) incoming DNS replies.
iptables was configured to block and log outgoing ICMP network traffic.
Reviewing the logs revealed ICMP port unreachable messages that were
destined to the legitimate DNS server, which were a good indication of a
DNS spoofing attack. Further to this evidence of a DNS spoofing attack,
since the DNS replies must come from port 53, analysis of the network
traffic using a packet dissector such as Ethereal revealed traffic that
looked like DNS replies apparently originating at the legitimate DNS
server.
--[ 8 - Conclusion
The RFC guidelines simplify DNS spoofing attacks against DNS client
resolvers since the attacker does not require information such as the IP
address of the potential victim's DNS server or the contents of DNS
queries sent by the potential victim. Microsoft Windows XP is more
susceptible to DNS spoofing attacks than Linux due to its poor
implementation of the RFC guidelines. Further simplifying DNS spoofing
attacks are Windows XP's inadequate matching of DNS requests to DNS
replies, and the predictable port number and transaction ID values -
behaviour that could be changed without violating the RFC guidelines.
Evidence of DNS spoofing attacks is minimised by the ability to disguise
DNS replies as NetBIOS traffic, the lack of configuration granularity and
traffic inspection of some firewalls, and Windows XP's failure to generate
ICMP error messages for excessive DNS replies.
RFC 791 (Internet Protocol) stating that a program must be "liberal in its
receiving behavior" and "must accept any datagram that it can interpret"
may have been acceptable in 1981 when the RFC was created and
interoperability was more important than security. However, the Internet
has changed from a somewhat trustworthy user base of representatives from
educational institutions and the US Department of Defense to now include
hackers and scammers, making security a high profile consideration.
Perhaps it is time for software based on this outdated perception of the
Internet to be changed as well.
The Internet community continues to wait for widespread adoption of
cryptographic digital signatures used to authenticate DNS transactions,
as discussed in RFC 2535 (Domain Name System Security Extensions). In the
meantime, the threat of DNS spoofing attacks could be minimised by
Microsoft improving the DNS implementation in all of their affected
operating systems. Such improvements include using random transaction ID
values, checking that the name in a DNS reply matches the name to be
resolved in the DNS request, and using a random source port for DNS
requests. These improvements would make attacks against DNS client
resolvers significantly more difficult to perform, and such improvements
would not violate the RFC guidelines.
|=----------------------------------------------------------------------=|
|=----------------------------------------------------------------------=|
########################################
# Injecting signals for Fun and Profit #
########################################
by shaun2k2 <shaunige@yahoo.co.uk>
--[ 1 - Introduction
More secure programming is on the rise, eliminating more generic program
exploitation vectors, such as stack-based overflows, heap overflows and symlink
bugs. Despite this, subtle vulnerabilities are often overlooked during code
audits, leaving so-called "secure" applications vulnerable to attack, but in a
less obvious manner. Secure design of signal-handlers is often not considered,
but I believe that this class of security holes deserves just as much attention
as more generic classes of bugs, such as buffer overflow bugs.
This paper intends to discuss problems faced when writing signal-handling
routines, how to exploit the problems, and presents ideas of how to avoid such
issues. A working knowledge of the C programming language and UNIX-like
operating systems would benefit the reader greatly, but is certainly not
essential.
--[ 2 - Signal Handling: An Overview
To understand what signal handlers are, one must first know what exactly a
signal is. In brief, signals are notifications delivered to a process to alert
the given process about "important" events concerning itself. For example,
users of an application can send signals using common keyboard Ctrl
combinations, such as Ctrl-C - which will send a SIGINT signal to the given
process.
Many different signals exist, but some of the more common (or useful) ones are:
SIGINT, SIGHUP, SIGKILL, SIGABRT, SIGTERM and SIGPIPE. Many more exist,
however. A list of available signals, according to the POSIX.1 standard,
can be found in the unix manual page signal(7).
It is worth noting that the signals SIGKILL and
SIGSTOP cannot be handled, ignored or blocked. Their 'action' can
not be changed.
"What are signal handlers", one might ask. The simple answer is that signal
handlers are small routines which are typically called when a pre-defined
signal, or set of signals, is delivered to the process it is running under
before the end of program execution - after execution flow has been directed to
a signal handling function, all instructions within the handler are executed in
turn. In larger applications, however, signal handling routines are often
written to complete a more complex set of tasks to ensure clean termination of
the program, such as; unlinking of tempory files, freeing of memory buffers,
appending log messages, and freeing file descriptors and/or sockets. Signal
handlers are generally defined as ordinary program functions, and are then
defined as the default handler for a certain signal usually near to the
beginning of the program.
Consider the sample program below:
--- sigint.c ---
#include <stdio.h>
#include <signal.h>
void sighndlr() {
printf("Ctrl-C caught!\n");
exit(0);
}
int main() {
signal(SIGINT, sighndlr);
while(1)
sleep(1);
/* should never reach here */
return(0);
}
--- EOF ---
'sigint.c' specifies that the function 'sighndlr' should be given control of
execution flow when a SIGINT signal is received by the program. The program
sleeps "forever", or until a SIGINT signal is received - in which case the
"Ctrl-C caught!" message is printed to the terminal - as seen below:
--- output ---
[root@localhost shaun]# gcc test.c -o test
[root@localhost shaun]# ./test
[... program sleeps ...]
Ctrl-C caught!
[root@localhost shaun]#
--- EOF ---
Generally speaking, a SIGINT signal is delivered when a user hits the Ctrl-C
combination at the keyboard, but a SIGINT signal can be generated by the
kill(1) utility.
However simple or complex the signal handler is, there are several potential
pitfalls which must be avoided during the development of the handler. Although
a signal handler may look "safe", problems may still arise, but may be
less-obvious to the unsuspecting eye. There are two main classes of problems
when dealing with signal-handler development - non-atomic process
modifications, and non-reentrant code, both of which are potentially critical
to system security.
--[ 3 - Non-atomic Modifications
Since signals can be delivered at almost any moment, and privileges often need
to be maintained (i.e root privileges in a SUID root application) for obvious
reasons (i.e for access to raw sockets, graphical resources, etc), signal
handling routines need to be written with extra care. If they are not, and
special privileges are held by the process at the particular time of signal
delivery, things could begin to go wrong very quickly. What is meant by
'non-atomic' is that the change in the program isn't permanant - it will
just be in place temporarily. To illustrate this, we will discuss a sample
vulnerable program.
Consider the following sample program:
--- atomicvuln.c ---
#include <stdio.h>
#include <signal.h>
void sighndlr() {
printf("Ctrl-C caught!\n");
printf("UID: %d\n", getuid());
/* other cleanup code... */
}
int showuid() {
printf("UID: %d\n", getuid());
return(0);
}
int main() {
int origuid = getuid();
signal(SIGINT, sighndlr);
setuid(0);
sleep(5);
setuid(origuid);
showuid();
return(0);
}
--- EOF ---
The above program should immediately spark up any security concious
programmer's paranoia, but the insecurity isn't immediately obvious
to everyone. As we can see from above, a signal handler is declared for
'SIGINT', and the program gives itself root privileges (so to speak). After
a delay of around five seconds, the privileges are revoked, and the
program is exited with success. However, if a SIGINT signal is received,
execution is directed to the SIGINT handler, 'sighdlr()'.
Let's look at some sample outputs:
--- output ---
[root@localhost shaun]# gcc test.c -o test
[root@localhost shaun]# chmod +s test
[root@localhost shaun]# exit
exit
[shaun@localhost shaun]$ ./test
[... program sleeps 5 seconds ...]
UID: 502
[shaun@localhost shaun]$ ./test
[... CTRL-C is typed ...]
Ctrl-C caught!
UID: 0
UID: 502
[shaun@localhost shaun]$
--- EOF ---
If you hadn't spotted the insecurity in 'atomicvuln.c' yet, the above output
should make things obvious; since the signal handling routine, 'sighdlr()', was
called when root privileges were still possessed, the friendly printf()
statements kindly tell us that our privileges are root (assuming the binary is
SUID root). And just to prove our theory, if we simply allow the program to
sleep for 5 seconds without sending an interrupt, the printf() statement kindly
tells us that our UID is 502 - my actual UID - as seen above.
With this, it is easy to understand where the flaw lies; if program execution
can be interrupted between the time when superuser privileges are given,
and the time when superuser privileges are revoked, the signal handling
code *will* be ran with root privileges. Just imagine - if the signal
handling routine included potentially sensitive code, compromisation of
root privileges could occur.
Although the sample program isn't an example of privilege escalation, it at
least demonstrates how non-atomic modifications can present security issues
when signal handling is involved. And do not assume that code similar to the
sample program above isn't found in popular security critical applications in
wide-spread use - it is. An example of vulnerable code similar to that of
above which is an application in wide-spread use, see [1] in the bibliography.
Non-reentrant Code
###################
Although it may not be obvious (and it's not), some glibc functions just
weren't designed to be reentered due to receipt of a signal, thus causing
potential problems for signal handlers which use them. An example of such a
function is the 'free()' function. According to 'free()'s man page, free()
"frees the memory space pointed to by ptr, which must have been
returned by a previous call to malloc(), calloc() or realloc(). Other-
wise, or if free(ptr) has already been called before, undefined
behaviour occurs. If ptr is NULL, no operation is performed."
As the man page snippet claims, free() can only be used to release memory which
was allocated using 'malloc()', else "undefined behavior" occurs. More
specifically, or in usual cases, the heap is corrupted, if free() is called on
a memory area which has already been free()d. Because of this implementation
design, reentrant signal routines which use 'free()' can be attacked.
Consider the below sample vulnerable program:
--- reentry.c ---
#include <stdio.h>
#include <signal.h>
#include <syslog.h>
#include <string.h>
#include <stdlib.h>
void *data1, *data2;
char *logdata;
void sighdlr() {
printf("Entered sighdlr()...\n");
syslog(LOG_NOTICE,"%s\n", logdata);
free(data2);
free(data1);
sleep(10);
exit(0);
}
int main(int argc, char *argv[]) {
logdata = argv[1];
data1 = strdup(argv[2]);
data2 = malloc(340);
signal(SIGHUP, sighdlr);
signal(SIGTERM, sighdlr);
sleep(10);
/* should never reach here */
return(0);
}
--- EOF ---
The above program defines a signal handler which frees allocated heap memory,
and sleeps for around 10 seconds. However, once the signal handler has been
entered, signals are not blocked, and thus can still be freely delivered. As
we learnt above, a duplicate call of free() on an already free()d memory area
will result in "undefined behavior" - possibly corruption of the heap memory.
As we can see, user-defined data is taken, and syslog() is also called fromo
the sig handler function - but how does syslog() work? 'syslog()' creates a
memory buffer stream, using two malloc() invokations - the first one allocates
a 'stream description structure', whilst the other creates a buffer suitable
for the actual syslog message data. This basis is essentially used to maintain
a tempory copy of the syslog message.
But why can this cause problems in context of co-usage of non-reentrant
routines? To find the answer, let's experiment a little, by attempting to
exploit the above program, which happens to be vulnerable.
--- output ---
[shaun@localhost shaun]$ ./test `perl -e 'print
"a"x100'` `perl -e 'print
"b"x410'` & sleep 1 ; killall -HUP test ; sleep 1 ;
killall -TERM test
[1] 2877
Entered sighdlr()...
Entered sighdlr()...
[1]+ Segmentation fault (core dumped) ./test
`perl -e 'print "a"x100'`
`perl -e 'print "b"x410'`
[shaun@localhost shaun]$ gdb -c core.2877
GNU gdb 5.2.1-2mdk (Mandrake Linux)
Copyright 2002 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General
Public License, and you are
welcome to change it and/or distribute copies of it
under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show
warranty" for details.
This GDB was configured as "i586-mandrake-linux-gnu".
Core was generated by `./test
aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa'.
Program terminated with signal 11, Segmentation fault.
#0 0x4008e9bb in ?? ()
(gdb) info reg
eax 0x61616161 1633771873
ecx 0x40138680 1075021440
edx 0x6965fa38 1768290872
ebx 0x4013c340 1075036992
esp 0xbfffeccc 0xbfffeccc
ebp 0xbfffed0c 0xbfffed0c
esi 0x80498d8 134519000
edi 0x61616160 1633771872
eip 0x4008e9bb 0x4008e9bb
eflags 0x10206 66054
cs 0x23 35
ss 0x2b 43
ds 0x2b 43
es 0x2b 43
fs 0x2b 43
gs 0x2b 43
fctrl 0x0 0
fstat 0x0 0
ftag 0x0 0
fiseg 0x0 0
fioff 0x0 0
foseg 0x0 0
fooff 0x0 0
---Type <return> to continue, or q <return> to quit---
fop 0x0 0
xmm0 {f = {0x0, 0x0, 0x0, 0x0}} {f = {0, 0, 0, 0}}
xmm1 {f = {0x0, 0x0, 0x0, 0x0}} {f = {0, 0, 0, 0}}
xmm2 {f = {0x0, 0x0, 0x0, 0x0}} {f = {0, 0, 0, 0}}
xmm3 {f = {0x0, 0x0, 0x0, 0x0}} {f = {0, 0, 0, 0}}
xmm4 {f = {0x0, 0x0, 0x0, 0x0}} {f = {0, 0, 0, 0}}
xmm5 {f = {0x0, 0x0, 0x0, 0x0}} {f = {0, 0, 0, 0}}
xmm6 {f = {0x0, 0x0, 0x0, 0x0}} {f = {0, 0, 0, 0}}
xmm7 {f = {0x0, 0x0, 0x0, 0x0}} {f = {0, 0, 0, 0}}
mxcsr 0x0 0
orig_eax 0xffffffff -1
(gdb) quit
[shaun@localhost shaun]$
--- EOF ---
Interesting. As we can see above, our large string of 'a's has found its way
into several program registers on stack - EAX and EDI. From this, we can
assume we are witnessing the "undefined behavior" we discussed earlier, when
the signal handler is reentered.
When the sample vulnerable program receives the second signal (SIGTERM), since
signals are not being ignored, the signal handler is reentered to handle this
second signal, causing something to go very wrong.
But why is this happening?
Since the second memory region (*data2) was free()d during the first entry of
the signal handler, syslog() re-uses this released memory for its own
purposes - storing its syslog message, because as the short syslog()
explanation above stated, two malloc() calls are present in most syslog()
implementations, and thus it re-uses the newly free()d memory - *data2.
After the usage of the memory once held as data2 by syslog(), a second
'free()' call is made on the memory region, because of reentry of the signal
handler function. As the free(3) man page stated, undefined behavior *will*
occur if the memory area was already free()d, and we happen to know that this
was the case. So when 'free()' was called again on *data2, free() landed
somewhere in the area containing the 'a's (hence 0x61 in hex), because
syslog() had re-used the freed area to store the syslog message, temporarily.
As the GDB output above illustrates, as long as user-input is used by
'syslog()' (and it is in this case), we have some control over the program
registers, when this "undefined behavior" (corruption of heap in most
cases) occurs. Because of this ability, exploitation is most likely a
possibility - it is left as an exercise to the reader to play with this sample
vulnerable program a little more, and determine if the vulnerability is
exploitable.
For the interested reader, 'free()' is not the only non-reentrant glibc
function. In general, it can be assumed that all glibc functions which are NOT
included within the following list are non-reentrant, and thus are not safe to
be used in signal handlers.
--
_exit(2), access(2), alarm(3), cfgetispeed(3), cfgetospeed(3),
cfsetispeed(3), cfsetospeed(3), chdir(2), chmod(2), chown(2),
close(2), creat(3), dup(2), dup2(2), execle(3), execve(2),
fcntl(2), fork(2), fpathconf(2), fstat(2), fsync(2), getegid(2),
geteuid(2), getgid(2), getgroups(2), getpgrp(2), getpid(2),
getppid(2), getuid(2), kill(2), link(2), lseek(2), mkdir(2),
mkfifo(2), open(2), pathconf(2), pause(3), pipe(2), raise(3),
read(2), rename(2), rmdir(2), setgid(2), setpgid(2), setsid(2),
setuid(2), sigaction(2), sigaddset(3), sigdelset(3),
sigemptyset(3), sigfillset(3), sigismember(3), signal(3),
sigpause(3), sigpending(2), sigprocmask(2), sigsuspend(2),
sleep(3), stat(2), sysconf(3), tcdrain(3), tcflow(3), tcflush(3),
tcgetattr(3), tcgetpgrp(3), tcsendbreak(3), tcsetattr(3),
tcsetpgrp(3), time(3), times(3), umask(2), uname(3), unlink(2),
utime(3), wait(2), waitpid(2), write(2)."
--
Secure Signal Handling
#######################
In general, signal handling vulnerabilities can be prevented by
--
1) Using only reentrant glibc functions within signal handlers -
This safe-guards against the possibility of "undefined behavior" or otherwise
as presented in the above example. However, this isn't *always* feasible,
especially when a programmers needs to accomplish tasks such as freeing
memory.
Other counter-measures, in this case, can protect against this. See below.
2) ignoring signals during signal handling routines -
As the obvious suggests, this programming practice will indefinately prevent
handling of signals during the execution of signal handling routines, thus
preventing signal handler reentry.
Consider the following signal handler template:
--- sighdlr.c ---
void sighdlr() {
signal(SIGINT, SIG_IGN);
signal(SIGABRT, SIG_IGN);
signal(SIGHUP, SIG_IGN);
/* ...ignore other signals ... */
/* cleanup code here */
exit(0);
}
--- EOF ---
As we can see above, signals are blocked before doing anything else in the
signal handling routine. This guarantees against signal handler reentry (or
almost does).
3) Ignoring signals whilst non-atomic process modifications are in place -
This involves blocking signals, in a similar way to the above code snippet,
during the execution of code with non-atomic modifications in place, such as
code execution with superuser privileges.
Consider the following code snippet:
--- nonatomicblock.c ---
/* code exec with non-atomic process modifications
starts here... */
signal(SIGINT, SIG_IGN);
signal(SIGABRT, SIG_IGN);
signal(SIGHUP, SIG_IGN);
/* block other signals if desired... */
setuid(0);
/* sensitive code here */
setuid(getuid());
/* sensitive code ends here */
signal(SIGINT, SIG_DFL);
signal(SIGABRT, SIG_DFL);
signal(SIGHUP, SIG_DFL);
/* ...code here... */
--- EOF ---
Before executing privileged code, signals are blocked. After execution of the
privileged code, privileges are dropped, and the signal action is set back to
the default action.
There are probably more ways of preventing signal vulnerabilities, but the
three above should be enough to implement semi-safe signal handlers.
Conclusion
###########
I hope this paper has at least touched upon possible problems encountered when
dealing with signals in C applications. If nothing else can be taken away from
this paper, my aim is to have outlined that secure programming practices should
always be applied when implementing signal handlers.
Full stop. Remember this.
If I have missed something out, given inaccurate information, or otherwise,
please feel free to drop me a line at the email address at the top of the
paper, providing your comments are nicely phrased.
Recommended reading is presented in the Bibliography below.
Bibliography
#############
Recommended reading material is:
--
"Delivering Signals for Fun and Profit" -
http://razor.bindview.com/publish/papers/signals.txt,
Michal Zalewski. Michal's
paper was a useful resource when writing this paper, and many ideas were gained
from this paper. Thanks Michal.
"Introduction To Unix Signals Programming" -
http://users.actcom.co.il/~choo/lupg/tutorials/signals/signals-programming.html,LUGPs.
"Procmail insecure signal handling vulnerability" -
http://xforce.iss.net/xforce/xfdb/6872
"Traceroute signal handling vulnerability" -
http://lwn.net/2000/1012/a/traceroute.php3
"signal(2) man page" -
http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc.cgi?coll=linux&db=man&fname=/usr/share/catman/man2/signal.2.html&srch=signal
"signal(7) man page" -
http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc.cgi?coll=linux&db=man&fname=/usr/share/catman/man7/signal.7.html&srch=signal
--
Greets
#######
Greets to:
--
Friends at HDC (or former HDC members), excluded.org,
#hackcanada, all @ GSO,
rider (happy be-lated birthday!).
All the other great people that I have met online.
--
Thanks guys.
Thank you for your time.
Shaun.
|=----------------------------------------------------------------------=|
|=----------------------------------------------------------------------=|
|=------------------=[ Pirating A Radio Station ]=----------------------=|
by j kuinga" <kuinga@hotmail.com>
At many Radio Stations to cut costs they now do what is called "central
casting." This is where many feeds are produced from one building and
handled by a group of engineers.
Why is this important? You could, disrupt the broadcast from the Central
Site, to the tower site, and ¡§create¡¨ your own programming, without the
hassles of buying a transmitter, getting the FCC licensing, and that type
of thing. We're showing you two different ways to have some fun--by
interrupting remote broadcasts, and by overtaking the radio station.
Radio Stations typically have ¡§Marti¡¦s¡¨ which are mini-transmitters, and
Marti Repeaters, typically in the 425-455 MHz Range. Some Ham Transmitters
will work in this range, and if not, check your local radio surplus store.
Marti¡¦s are typically used to rebroadcast High School Football and
basketball games, as well as commercial "live events" and it¡¦s something as
simple as over-powering the signal, in order to get your message through.
Be forewarned, there typically is a live person on the other end of that
transmitter¡Xthey¡¦re probably not paying attention, because they¡¦re
getting paid $5.50/hour¡Xbut, they have they ability to turn you off.
How to find the frequency? Well, you could always SE the engineer at the
station and ask, however, most of them are grumpy old radio buffs, so you
might not get anywhere. I suggest a good copy of ¡§Police Call,¡¨ which has
a LOT of frequencies in there for things like radio stations.
I use a home-made setup for finding particular frequencies out. Having some
essential tools like a good, directional antenna, frequency counter, and
very accurate transmitter, along with breadboard and essential components,
typically are common in finding what you need to know. I also drive a Big
White Van, complete with Mast and Bucket, so I can optimally 'place' the
antenna at the right height and direction, that I obtained at a school
auction for reallly cheap. (e.g., under $500, even had 18" racks in it and a
nice generator)
Most Radio Stations doing this have what they call a ¡§STL,¡¨ or Studio to
Transmitter Link. This is typically in the 800 or 900 Mhz range, and the
same, general ideas apply. You find the general direction in which the
antenna is pointed, then you overpower the signal. Since you
(idealistically) would be within a few miles of the transmitter, not 30 or
50 miles like the Central-Casting spot, you would overpower the transmitter,
and start your own pirate radio station. Most stations however, have an
¡§Air¡¨ monitor, and can turn the remote transmitter off by pressing a
button on their STL. However, if you¡¦re closer to it, you¡¦ve got control
until the station engineer comes down to manually pull the plug on your
transmitter.
If you see black vans with antennas and they look like they're doing sweeps,
chances are, they're either a) with the audit crew of the local cable
company, or b) looking for your ass.
kuinga@hotmail.com
|=[ EOF ]=---------------------------------------------------------------=|
TUCoPS is optimized to look best in Firefox® on a widescreen monitor (1440x900 or better).
Site design & layout copyright © 1986-2024 AOH
