RFC 6056
Recommendations for Transport-Protocol Port Randomization
Internet Engineering Task Force (IETF) M. Larsen Request for Comments: 6056 Tieto BCP: 156 F. Gont Category: Best Current Practice UTN/FRH ISSN: 2070-1721 January 2011 Recommendations for Transport-Protocol Port RandomizationAbstract
During the last few years, awareness has been raised about a number of "blind" attacks that can be performed against the Transmission Control Protocol (TCP) and similar protocols. The consequences of these attacks range from throughput reduction to broken connections or data corruption. These attacks rely on the attacker's ability to guess or know the five-tuple (Protocol, Source Address, Destination Address, Source Port, Destination Port) that identifies the transport protocol instance to be attacked. This document describes a number of simple and efficient methods for the selection of the client port number, such that the possibility of an attacker guessing the exact value is reduced. While this is not a replacement for cryptographic methods for protecting the transport-protocol instance, the aforementioned port selection algorithms provide improved security with very little effort and without any key management overhead. The algorithms described in this document are local policies that may be incrementally deployed and that do not violate the specifications of any of the transport protocols that may benefit from them, such as TCP, UDP, UDP-lite, Stream Control Transmission Protocol (SCTP), Datagram Congestion Control Protocol (DCCP), and RTP (provided that the RTP application explicitly signals the RTP and RTCP port numbers). Status of This Memo This memo documents an Internet Best Current Practice. This document is a product of the Internet Engineering Task Force (IETF). It represents the consensus of the IETF community. It has received public review and has been approved for publication by the Internet Engineering Steering Group (IESG). Further information on BCPs is available in Section 2 of RFC 5741. Information about the current status of this document, any errata, and how to provide feedback on it may be obtained at http://www.rfc-editor.org/info/rfc6056.
Copyright Notice Copyright (c) 2011 IETF Trust and the persons identified as the document authors. All rights reserved. This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License. This document may contain material from IETF Documents or IETF Contributions published or made publicly available before November 10, 2008. The person(s) controlling the copyright in some of this material may not have granted the IETF Trust the right to allow modifications of such material outside the IETF Standards Process. Without obtaining an adequate license from the person(s) controlling the copyright in such materials, this document may not be modified outside the IETF Standards Process, and derivative works of it may not be created outside the IETF Standards Process, except to format it for publication as an RFC or to translate it into languages other than English.
Table of Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 4 2. Ephemeral Ports . . . . . . . . . . . . . . . . . . . . . . . 5 2.1. Traditional Ephemeral Port Range . . . . . . . . . . . . . 5 2.2. Ephemeral Port Selection . . . . . . . . . . . . . . . . . 6 2.3. Collision of instance-ids . . . . . . . . . . . . . . . . 7 3. Obfuscating the Ephemeral Port Selection . . . . . . . . . . . 8 3.1. Characteristics of a Good Algorithm for the Obfuscation of the Ephemeral Port Selection . . . . . . . 8 3.2. Ephemeral Port Number Range . . . . . . . . . . . . . . . 10 3.3. Algorithms for the Obfuscation of the Ephemeral Port Selection . . . . . . . . . . . . . . . . . . . . . . . . 11 3.3.1. Algorithm 1: Simple Port Randomization Algorithm . . . 11 3.3.2. Algorithm 2: Another Simple Port Randomization Algorithm . . . . . . . . . . . . . . . . . . . . . . 13 3.3.3. Algorithm 3: Simple Hash-Based Port Selection Algorithm . . . . . . . . . . . . . . . . . . . . . . 14 3.3.4. Algorithm 4: Double-Hash Port Selection Algorithm . . 16 3.3.5. Algorithm 5: Random-Increments Port Selection Algorithm . . . . . . . . . . . . . . . . . . . . . . 18 3.4. Secret-Key Considerations for Hash-Based Port Selection Algorithms . . . . . . . . . . . . . . . . . . . 19 3.5. Choosing an Ephemeral Port Selection Algorithm . . . . . . 20 4. Interaction with Network Address Port Translation (NAPT) . . . 22 5. Security Considerations . . . . . . . . . . . . . . . . . . . 23 6. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 24 7. References . . . . . . . . . . . . . . . . . . . . . . . . . . 24 7.1. Normative References . . . . . . . . . . . . . . . . . . . 24 7.2. Informative References . . . . . . . . . . . . . . . . . . 25 Appendix A. Survey of the Algorithms in Use by Some Popular Implementations . . . . . . . . . . . . . . . . . . . 28 A.1. FreeBSD . . . . . . . . . . . . . . . . . . . . . . . . . 28 A.2. Linux . . . . . . . . . . . . . . . . . . . . . . . . . . 28 A.3. NetBSD . . . . . . . . . . . . . . . . . . . . . . . . . . 28 A.4. OpenBSD . . . . . . . . . . . . . . . . . . . . . . . . . 28 A.5. OpenSolaris . . . . . . . . . . . . . . . . . . . . . . . 28
1. Introduction
Recently, awareness has been raised about a number of "blind" attacks (i.e., attacks that can be performed without the need to sniff the packets that correspond to the transport protocol instance to be attacked) that can be performed against the Transmission Control Protocol (TCP) [RFC0793] and similar protocols. The consequences of these attacks range from throughput reduction to broken connections or data corruption [RFC5927] [RFC4953] [Watson]. All these attacks rely on the attacker's ability to guess or know the five-tuple (Protocol, Source Address, Source port, Destination Address, Destination Port) that identifies the transport protocol instance to be attacked. Services are usually located at fixed, "well-known" ports [IANA] at the host supplying the service (the server). Client applications connecting to any such service will contact the server by specifying the server IP address and service port number. The IP address and port number of the client are normally left unspecified by the client application and thus are chosen automatically by the client networking stack. Ports chosen automatically by the networking stack are known as ephemeral ports [Stevens]. While the server IP address, the well-known port, and the client IP address may be known by an attacker, the ephemeral port of the client is usually unknown and must be guessed. This document describes a number of algorithms for the selection of ephemeral port numbers, such that the possibility of an off-path attacker guessing the exact value is reduced. They are not a replacement for cryptographic methods of protecting a transport- protocol instance such as IPsec [RFC4301], the TCP MD5 signature option [RFC2385], or the TCP Authentication Option [RFC5925]. For example, they do not provide any mitigation in those scenarios in which the attacker is able to sniff the packets that correspond to the transport protocol instance to be attacked. However, the proposed algorithms provide improved resistance to off-path attacks with very little effort and without any key management overhead. The mechanisms described in this document are local modifications that may be incrementally deployed, and that do not violate the specifications of any of the transport protocols that may benefit from them, such as TCP [RFC0793], UDP [RFC0768], SCTP [RFC4960], DCCP [RFC4340], UDP-lite [RFC3828], and RTP [RFC3550] (provided the RTP application explicitly signals the RTP and RTCP port numbers with, e.g., [RFC3605]).
Since these mechanisms are obfuscation techniques, focus has been on
a reasonable compromise between the level of obfuscation and the ease
of implementation. Thus, the algorithms must be computationally
efficient and not require substantial state.
We note that while the technique of mitigating "blind" attacks by
obfuscating the ephemeral port selection is well-known as "port
randomization", the goal of the algorithms described in this document
is to reduce the chances of an attacker guessing the ephemeral ports
selected for new transport protocol instances, rather than to
actually produce mathematically random sequences of ephemeral ports.
Throughout this document, we will use the term "transport-protocol
instance" as a general term to refer to an instantiation of a
transport protocol (e.g., a "connection" in the case of connection-
oriented transport protocols) and the term "instance-id" as a short-
handle to refer to the group of values that identify a transport-
protocol instance (e.g., in the case of TCP, the five-tuple
{Protocol, IP Source Address, TCP Source Port, IP Destination
Address, TCP Destination Port}).
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119 [RFC2119].
2. Ephemeral Ports
2.1. Traditional Ephemeral Port Range
The Internet Assigned Numbers Authority (IANA) assigns the unique
parameters and values used in protocols developed by the Internet
Engineering Task Force (IETF), including well-known ports [IANA].
IANA has reserved the following use of the 16-bit port range of TCP
and UDP:
o The Well-Known Ports, 0 through 1023.
o The Registered Ports, 1024 through 49151
o The Dynamic and/or Private Ports, 49152 through 65535
The dynamic port range defined by IANA consists of the 49152-65535
range, and is meant for the selection of ephemeral ports.
2.2. Ephemeral Port Selection
As each communication instance is identified by the five-tuple {protocol, local IP address, local port, remote IP address, remote port}, the selection of ephemeral port numbers must result in a unique five-tuple. Selection of ephemeral ports such that they result in unique instance-ids (five-tuples) is handled by some implementations by having a per-protocol global "next_ephemeral" variable that is equal to the previously chosen ephemeral port + 1, i.e., the selection process is: /* Initialization at system boot time. Could be random */ next_ephemeral = min_ephemeral; /* Ephemeral port selection function */ count = max_ephemeral - min_ephemeral + 1; do { port = next_ephemeral; if (next_ephemeral == max_ephemeral) { next_ephemeral = min_ephemeral; } else { next_ephemeral++; } if (check_suitable_port(port)) return port; count--; } while (count > 0); return ERROR; Traditional BSD Port Selection Algorithm Note: check_suitable_port() is a function that checks whether the resulting port number is acceptable as an ephemeral port. That is, it checks whether the resulting port number is unique and may, in addition, check that the port number is not in use for a connection in the LISTEN or CLOSED states and that the port number is not in the list of port numbers that should not be allocated as ephemeral ports. In BSD-derived systems, the check_suitable_port() would correspond to the in_pcblookup_local() function, where all the necessary checks would be performed.
This algorithm works adequately provided that the number of transport-protocol instances (for each transport protocol) that have a lifetime longer than it takes to exhaust the total ephemeral port range is small, so that collisions of instance-ids are rare. However, this method has the drawback that the "next_ephemeral" variable and thus the ephemeral port range is shared between all transport-protocol instances, and the next ports chosen by the client are easy to predict. If an attacker operates an "innocent" server to which the client connects, it is easy to obtain a reference point for the current value of the "next_ephemeral" variable. Additionally, if an attacker could force a client to periodically establish, e.g., a new TCP connection to an attacker-controlled machine (or through an attacker-observable path), the attacker could subtract consecutive source port values to obtain the number of outgoing TCP connections established globally by the target host within that time period (up to wrap-around issues and instance-id collisions, of course).2.3. Collision of instance-ids
While it is possible for the ephemeral port selection algorithm to verify that the selected port number results in a instance-id that is not currently in use by that system, the resulting five-tuple may still be in use at a remote system. For example, consider a scenario in which a client establishes a TCP connection with a remote web server, and the web server performs the active close on the connection. While the state information for this connection will disappear at the client side (that is, the connection will be moved to the fictional CLOSED state), the instance-id will remain in the TIME-WAIT state at the web server for 2*MSL (Maximum Segment Lifetime). If the same client tried to create a new incarnation of the previous connection (that is, a connection with the same instance-id as the one in the TIME_WAIT state at the server), an instance-id "collision" would occur. The effect of these collisions range from connection-establishment failures to TIME-WAIT state assassination (with the potential of data corruption) [RFC1337]. In scenarios in which a specific client establishes TCP connections with a specific service at a server, these problems become evident. Therefore, an ephemeral port selection algorithm should ideally minimize the rate of instance-id collisions. A simple approach to minimize the rate of these collisions would be to choose port numbers incrementally, so that a given port number would not be reused until the rest of the port numbers in the ephemeral port range have been used for a transport protocol instance. However, if a single global variable were used to keep track of the last ephemeral port selected, ephemeral port numbers would be trivially predictable, thus making it easier for an off-path
attacker to "guess" the instance-id in use by a target transport- protocol instance. Sections 3.3.3 and 3.3.4 describe algorithms that select port numbers incrementally, while still making it difficult for an off-path attacker to predict the ephemeral ports used for future transport-protocol instances. A simple but inefficient approach to minimize the rate of collisions of instance-ids would be, e.g., in the case of TCP, for both endpoints of a TCP connection to keep state about recent connections (e.g., have both endpoints end up in the TIME-WAIT state).3. Obfuscating the Ephemeral Port Selection
3.1. Characteristics of a Good Algorithm for the Obfuscation of the Ephemeral Port Selection
There are several factors to consider when designing an algorithm for selecting ephemeral ports, which include: o Minimizing the predictability of the ephemeral port numbers used for future transport-protocol instances. o Minimizing collisions of instance-ids. o Avoiding conflict with applications that depend on the use of specific port numbers. Given the goal of improving the transport protocol's resistance to attack by obfuscation of the instance-id selection, it is key to minimize the predictability of the ephemeral ports that will be selected for new transport-protocol instances. While the obvious approach to address this requirement would be to select the ephemeral ports by simply picking a random value within the chosen port number range, this straightforward policy may lead to collisions of instance-ids, which could lead to the interoperability problems (e.g., delays in the establishment of new connections, failures in connection establishment, or data corruption) discussed in Section 2.3. As discussed in Section 1, it is worth noting that while the technique of mitigating "blind" attacks by obfuscating the ephemeral port selection is well-known as "port randomization", the goal of the algorithms described in this document is to reduce the chances that an attacker will guess the ephemeral ports selected for new transport-protocol instances, rather than to actually produce sequences of mathematically random ephemeral port numbers.
It is also worth noting that, provided adequate algorithms are in use, the larger the range from which ephemeral ports are selected, the smaller the chances of an attacker are to guess the selected port number. In scenarios in which a specific client establishes transport- protocol instances with a specific service at a server, the problems described in Section 2.3 become evident. A good algorithm to minimize the collisions of instance-ids would consider the time a given five-tuple was last used, and would avoid reusing the last recently used five-tuples. A simple approach to minimize the rate of collisions would be to choose port numbers incrementally, so that a given port number would not be reused until the rest of the port numbers in the ephemeral port range have been used for a transport- protocol instance. However, if a single global variable were used to keep track of the last ephemeral port selected, ephemeral port numbers would be trivially predictable. It is important to note that a number of applications rely on binding specific port numbers that may be within the ephemeral port range. If such an application were run while the corresponding port number were in use, the application would fail. Therefore, ephemeral port selection algorithms avoid using those port numbers. Port numbers that are currently in use by a TCP in the LISTEN state should not be allowed for use as ephemeral ports. If this rule is not complied with, an attacker could potentially "steal" an incoming connection to a local server application in at least two different ways. Firstly, an attacker could issue a connection request to the victim client at roughly the same time the client tries to connect to the victim server application [CPNI-TCP] [TCP-SEC]. If the SYN segment corresponding to the attacker's connection request and the SYN segment corresponding to the victim client "cross each other in the network", and provided the attacker is able to know or guess the ephemeral port used by the client, a TCP "simultaneous open" scenario would take place, and the incoming connection request sent by the client would be matched with the attacker's socket rather than with the victim server application's socket. Secondly, an attacker could specify a more specific socket than the "victim" socket (e.g., specify both the local IP address and the local TCP port), and thus incoming SYN segments matching the attacker's socket would be delivered to the attacker, rather than to the "victim" socket (see Section 10.1 of [CPNI-TCP]). It should be noted that most applications based on popular implementations of the TCP API (such as the Sockets API) perform "passive opens" in three steps. Firstly, the application obtains a file descriptor to be used for inter-process communication (e.g., by
issuing a socket() call). Secondly, the application binds the file descriptor to a local TCP port number (e.g., by issuing a bind() call), thus creating a TCP in the fictional CLOSED state. Thirdly, the aforementioned TCP is put in the LISTEN state (e.g., by issuing a listen() call). As a result, with such an implementation of the TCP API, even if port numbers in use for TCPs in the LISTEN state were not allowed for use as ephemeral ports, there is a window of time between the second and the third steps in which an attacker could be allowed to select a port number that would be later used for listening to incoming connections. Therefore, these implementations of the TCP API should enforce a stricter requirement for the allocation of port numbers: port numbers that are in use by a TCP in the LISTEN or CLOSED states should not be allowed for allocation as ephemeral ports [CPNI-TCP] [TCP-SEC]. The aforementioned issue does not affect SCTP, since most SCTP implementations do not allow a socket to be bound to the same port number unless a specific socket option (SCTP_REUSE_PORT) is issued on the socket (i.e., this behavior needs to be explicitly allowed beforehand). An example of a typical SCTP socket API can be found in [SCTP-SOCKET]. DCCP is not affected by the exploitation of "simultaneous opens" to "steal" incoming connections, as the server and the client state machines are different [RFC4340]. However, it may be affected by the vector involving binding a more specific socket. As a result, those tuples {local IP address, local port, Service Code} that are in use by a local socket should not be allowed for allocation as ephemeral ports.3.2. Ephemeral Port Number Range
As mentioned in Section 2.1, the dynamic ports consist of the range 49152-65535. However, ephemeral port selection algorithms should use the whole range 1024-65535. This range includes the IANA Registered Ports; thus, some of these port numbers may be needed for providing a particular service at the local host, which could result in the problems discussed in Section 3.1. As a result, port numbers that may be needed for providing a particular service at the local host SHOULD NOT be included in the pool of port numbers available for ephemeral port randomization. If the host does not provide a particular service, the port can be safely allocated to ordinary processes. A possible workaround for this potential problem would be to maintain a local list of the port numbers that should not be allocated as ephemeral ports. Thus, before allocating a port number, the
ephemeral port selection function would check this list, avoiding the allocation of ports that may be needed for specific applications. Rather than naively excluding all the registered ports, administrators should identify services that may be offered by the local host and SHOULD exclude only the corresponding registered ports. Ephemeral port selection algorithms SHOULD use the largest possible port range, since this reduces the chances of an off-path attacker of guessing the selected port numbers.3.3. Algorithms for the Obfuscation of the Ephemeral Port Selection
Ephemeral port selection algorithms SHOULD obfuscate the selection of their ephemeral ports, since this helps to mitigate a number of attacks that depend on the attacker's ability to guess or know the five-tuple that identifies the transport-protocol instance to be attacked. The following subsections describe a number of algorithms that could be implemented in order to obfuscate the selection of ephemeral port numbers.3.3.1. Algorithm 1: Simple Port Randomization Algorithm
In order to address the security issues discussed in Sections 1 and 2.2, a number of systems have implemented simple ephemeral port number randomization, as follows:
/* Ephemeral port selection function */
num_ephemeral = max_ephemeral - min_ephemeral + 1;
next_ephemeral = min_ephemeral + (random() % num_ephemeral);
count = num_ephemeral;
do {
if(check_suitable_port(port))
return next_ephemeral;
if (next_ephemeral == max_ephemeral) {
next_ephemeral = min_ephemeral;
} else {
next_ephemeral++;
}
count--;
} while (count > 0);
return ERROR;
Algorithm 1
Note:
random() is a function that returns a 32-bit pseudo-random
unsigned integer number. Note that the output needs to be
unpredictable, and typical implementations of POSIX random()
function do not necessarily meet this requirement. See [RFC4086]
for randomness requirements for security.
All the variables (in this and all the algorithms discussed in
this document) are unsigned integers.
Since the initially chosen port may already be in use with IP
addresses and server port that are identical to the ones being used
for the socket for which the ephemeral port is to be selected, the
resulting five-tuple might not be unique. Therefore, multiple ports
may have to be tried and verified against all existing transport-
protocol instances before a port can be chosen.
Web proxy servers, Network Address Port Translators (NAPTs)
[RFC2663], and other middleboxes aggregate multiple peers into the
same port space and thus increase the population of used ephemeral
ports, and hence the chances of collisions of instance-ids. However,
[Allman] has shown that at least in the network scenarios used for
measuring the collision properties of the algorithms described in
this document, the collision rate resulting from the use of the
aforementioned middleboxes is nevertheless very low.
Since this algorithm performs port selection without taking into account the port numbers previously chosen, it has the potential of reusing port numbers too quickly, thus possibly leading to collisions of instance-ids. Even if a given instance-id is verified to be unique by the port selection algorithm, the instance-id might still be in use at the remote system. In such a scenario, a connection request could possibly fail ([Silbersack] describes this problem for the TCP case). However, this algorithm is biased towards the first available port after a sequence of unavailable port numbers. If the local list of registered port numbers that should not be allocated as ephemeral ports (as described in Section 3.2) is significant, an attacker may actually have a significantly better chance of guessing a port number. This algorithm selects ephemeral port numbers randomly and thus reduces the chances that an attacker will guess the ephemeral port selected for a target transport-protocol instance. Additionally, it prevents attackers from obtaining the number of outgoing transport- protocol instances (e.g., TCP connections) established by the client in some period of time.3.3.2. Algorithm 2: Another Simple Port Randomization Algorithm
The following pseudo-code illustrates another algorithm for selecting a random port number, in which in the event a local instance-id collision is detected, another port number is selected randomly: /* Ephemeral port selection function */ num_ephemeral = max_ephemeral - min_ephemeral + 1; next_ephemeral = min_ephemeral + (random() % num_ephemeral); count = num_ephemeral; do { if(check_suitable_port(port)) return next_ephemeral; next_ephemeral = min_ephemeral + (random() % num_ephemeral); count--; } while (count > 0); return ERROR; Algorithm 2
When there are a large number of port numbers already in use for the same destination endpoint, this algorithm might be unable (with a very small remaining probability) to select an ephemeral port (i.e., it would return "ERROR"), even if there are still a few port numbers available that would result in unique five-tuples. However, the results in [Allman] have shown that in common scenarios, one port choice is enough, and in most cases where more than one choice is needed, two choices suffice. Therefore, in those scenarios this would not be problem.3.3.3. Algorithm 3: Simple Hash-Based Port Selection Algorithm
We would like to achieve the port-reuse properties of the traditional BSD port selection algorithm (described in Section 2.2), while at the same time achieve the unpredictability properties of Algorithm 1 and Algorithm 2. Ideally, we would like a "next_ephemeral" value for each set of (local IP address, remote IP addresses, remote port), so that the port-reuse frequency is the lowest possible. Each of these "next_ephemeral" variables should be initialized with random values within the ephemeral port range and, together, these would thus separate the ephemeral port space of the transport-protocol instances on a "per-destination endpoint" basis (this "separation of the ephemeral port space" means that transport-protocol instances with different remote endpoints will not have different sequences of port numbers, i.e., will not be part of the same ephemeral port sequence as in the case of the traditional BSD ephemeral port selection algorithm). Since we do not want to maintain in memory all these "next_ephemeral" values, we propose an offset function F() that can be computed from the local IP address, remote IP address, remote port, and a secret key. F() will yield (practically) different values for each set of arguments, i.e.:
/* Initialization at system boot time. Could be random. */
next_ephemeral = 0;
/* Ephemeral port selection function */
num_ephemeral = max_ephemeral - min_ephemeral + 1;
offset = F(local_IP, remote_IP, remote_port, secret_key);
count = num_ephemeral;
do {
port = min_ephemeral +
(next_ephemeral + offset) % num_ephemeral;
next_ephemeral++;
if(check_suitable_port(port))
return port;
count--;
} while (count > 0);
return ERROR;
Algorithm 3
In other words, the function F() provides a "per-destination
endpoint" fixed offset within the global ephemeral port range. Both
the "offset" and "next_ephemeral" variables may take any value within
the storage type range since we are restricting the resulting port in
a similar way as in Algorithm 1 (described in Section 3.3.1). This
allows us to simply increment the "next_ephemeral" variable and rely
on the unsigned integer to wrap around.
The function F() should be a cryptographic hash function like MD5
[RFC1321]. The function should use both IP addresses, the remote
port, and a secret key value to compute the offset. The remote IP
address is the primary separator and must be included in the offset
calculation. The local IP address and remote port may in some cases
be constant and thus not improve the ephemeral port space separation;
however, they should also be included in the offset calculation.
Cryptographic algorithms stronger than, e.g., MD5 should not be
necessary, given that Algorithm 3 is simply a technique for the
obfuscation of the selection of ephemeral ports. The secret should
be chosen to be as random as possible (see [RFC4086] for
recommendations on choosing secrets).
Note that on multiuser systems, the function F() could include user-
specific information, thereby providing protection not only on a
host-to-host basis, but on a user to service basis. In fact, any
identifier of the remote entity could be used, depending on
availability and the granularity requested. With SCTP, both
hostnames and alternative IP addresses may be included in the
association negotiation, and either of these could be used in the
offset function F().
When multiple unique identifiers are available, any of these can be
chosen as input to the offset function F() since they all uniquely
identify the remote entity. However, in cases like SCTP where the
ephemeral port must be unique across all IP address permutations, we
should ideally always use the same IP address to get a single
starting offset for each association negotiation with a given remote
entity to minimize the possibility of collisions. A simple numerical
sorting of the IP addresses and always using the numerically lowest
could achieve this. However, since most protocols will generally
report the same IP addresses in the same order in each association
setup, this sorting is most likely not necessary and the "first one"
can simply be used.
The ability of hostnames to uniquely define hosts can be discussed,
and since SCTP always includes at least one IP address, we recommend
using this as input to the offset function F() and ignoring hostname
chunks when searching for ephemeral ports.
It should be noted that, as this algorithm uses a global counter
("next_ephemeral") for selecting ephemeral ports, if an attacker
could, e.g., force a client to periodically establish a new TCP
connection to an attacker-controlled machine (or through an attacker-
observable path), the attacker could subtract consecutive source port
values to obtain the number of outgoing TCP connections established
globally by the target host within that time period (up to wrap-
around issues and five-tuple collisions, of course).
3.3.4. Algorithm 4: Double-Hash Port Selection Algorithm
A trade-off between maintaining a single global "next_ephemeral"
variable and maintaining 2**N "next_ephemeral" variables (where N is
the width of the result of F()) could be achieved as follows. The
system would keep an array of TABLE_LENGTH short integers, which
would provide a separation of the increment of the "next_ephemeral"
variable. This improvement could be incorporated into Algorithm 3 as
follows:
/* Initialization at system boot time */
for(i = 0; i < TABLE_LENGTH; i++)
table[i] = random() % 65536;
/* Ephemeral port selection function */
num_ephemeral = max_ephemeral - min_ephemeral + 1;
offset = F(local_IP, remote_IP, remote_port, secret_key1);
index = G(local_IP, remote_IP, remote_port, secret_key2);
count = num_ephemeral;
do {
port = min_ephemeral + (offset + table[index]) % num_ephemeral;
table[index]++;
if(check_suitable_port(port))
return port;
count--;
} while (count > 0);
return ERROR;
Algorithm 4
"table[]" could be initialized with mathematically random values, as
indicated by the initialization code in pseudo-code above. The
function G() should be a cryptographic hash function like MD5
[RFC1321]. It should use both IP addresses, the remote port, and a
secret key value to compute a value between 0 and (TABLE_LENGTH-1).
Alternatively, G() could take an "offset" as input, and perform the
exclusive-or (XOR) operation between all the bytes in "offset".
The array "table[]" assures that successive transport-protocol
instances with the same remote endpoint will use increasing ephemeral
port numbers. However, incrementation of the port numbers is
separated into TABLE_LENGTH different spaces, and thus the port-reuse
frequency will be (probabilistically) lower than that of Algorithm 3.
That is, a new transport-protocol instance with some remote endpoint
will not necessarily cause the "next_ephemeral" variable
corresponding to other endpoints to be incremented.
It is interesting to note that the size of "table[]" does not limit
the number of different port sequences, but rather separates the
*increments* into TABLE_LENGTH different spaces. The port sequence
will result from adding the corresponding entry of "table[]" to the
variable "offset", which selects the actual port sequence (as in
Algorithm 3). [Allman] has found that a TABLE_LENGTH of 10 can
result in an improvement over Algorithm 3. Further increasing the TABLE_LENGTH will increase the unpredictability of the resulting port number, and possibly further decrease the collision rate. An attacker can perform traffic analysis for any "increment space" into which the attacker has "visibility" -- namely, the attacker can force the client to establish a transport-protocol instance whose G(offset) identifies the target "increment space". However, the attacker's ability to perform traffic analysis is very reduced when compared to the traditional BSD algorithm (described in Section 2.2) and Algorithm 3. Additionally, an implementation can further limit the attacker's ability to perform traffic analysis by further separating the increment space (that is, using a larger value for TABLE_LENGTH).3.3.5. Algorithm 5: Random-Increments Port Selection Algorithm
[Allman] introduced another port selection algorithm, which offers a middle ground between the algorithms that select ephemeral ports independently at random (such as those described in Sections 3.3.1 and 3.3.2), and those that offer obfuscation with less randomization (such as those described in Sections 3.3.3 and 3.3.4). /* Initialization code at system boot time. */ next_ephemeral = random() % 65536; /* Initialization value */ N = 500; /* Determines the trade-off */ /* Ephemeral port selection function */ num_ephemeral = max_ephemeral - min_ephemeral + 1; count = num_ephemeral; do { next_ephemeral = next_ephemeral + (random() % N) + 1; port = min_ephemeral + (next_ephemeral % num_ephemeral); if(check_suitable_port(port)) return port; count--; } while (count > 0); return ERROR; Algorithm 5
This algorithm aims at producing a monotonically increasing sequence to prevent the collision of instance-ids, while avoiding the use of fixed increments, which would lead to trivially predictable sequences. The value "N" allows for direct control of the trade-off between the level of unpredictability and the port-reuse frequency. The smaller the value of "N", the more similar this algorithm is to the traditional BSD port selection algorithm (described in Section 2.2). The larger the value of "N", the more similar this algorithm is to the algorithm described in Section 3.3.1 of this document. When the port numbers wrap, there is the risk of collisions of instance-ids. Therefore, "N" should be selected according to the following criteria: o It should maximize the wrapping time of the ephemeral port space. o It should minimize collisions of instance-ids. o It should maximize the unpredictability of selected port numbers. Clearly, these are competing goals, and the decision of which value of "N" to use is a trade-off. Therefore, the value of "N" should be configurable so that system administrators can make the trade-off for themselves.3.4. Secret-Key Considerations for Hash-Based Port Selection Algorithms
Every complex manipulation (like MD5) is no more secure than the input values, and in the case of ephemeral ports, the secret key. If an attacker is aware of which cryptographic hash function is being used by the victim (which we should expect), and the attacker can obtain enough material (e.g., ephemeral ports chosen by the victim), the attacker may simply search the entire secret-key space to find matches. To protect against this, the secret key should be of a reasonable length. Key lengths of 128 bits should be adequate. Another possible mechanism for protecting the secret key is to change it after some time. If the host platform is capable of producing reasonably good random data, the secret key can be changed automatically. Changing the secret will cause abrupt shifts in the chosen ephemeral ports, and consequently collisions may occur. That is, upon changing the secret, the "offset" value (see Sections 3.3.3 and 3.3.4) used
for each destination endpoint will be different from that computed
with the previous secret, thus leading to the selection of a port
number recently used for connecting to the same endpoint.
Thus, the change in secret key should be done with consideration and
could be performed whenever one of the following events occur:
o The system is being bootstrapped.
o Some predefined/random time has expired.
o The secret key has been used sufficiently often that it should be
regarded as insecure now.
o There are few active transport-protocol instances (i.e.,
possibility of a collision is low).
o System load is low (i.e., the performance overhead of local
collisions is tolerated).
o There is enough random data available to change the secret key
(pseudo-random changes should not be done).
3.5. Choosing an Ephemeral Port Selection Algorithm
[Allman] is an empirical study of the properties of the algorithms
described in this document, which has found that all the algorithms
described in this document offer low collision rates -- at most 0.3%.
That is, in those network scenarios assessed by [Allman], all of the
algorithms described in this document perform well in terms of
collisions of instance-ids. However, these results may vary
depending on the characteristics of network traffic and the specific
network setup.
The algorithm described in Section 2.2 is the traditional ephemeral
port selection algorithm implemented in BSD-derived systems. It
generates a global sequence of ephemeral port numbers, which makes it
trivial for an attacker to predict the port number that will be used
for a future transport protocol instance. However, it is very simple
and leads to a low port-reuse frequency.
Algorithm 1 and Algorithm 2 have the advantage that they provide
actual randomization of the ephemeral ports. However, they may
increase the chances of port number collisions, which could lead to
the failure of a connection establishment attempt. [Allman] found
that these two algorithms show the largest collision rates (among all
the algorithms described in this document).
Algorithm 3 provides complete separation in local and remote IP addresses and remote port space, and only limited separation in other dimensions (see Section 3.4). However, implementations should consider the performance impact of computing the cryptographic hash used for the offset. Algorithm 4 improves Algorithm 3, usually leading to a lower port- reuse frequency, at the expense of more processor cycles used for computing G(), and additional kernel memory for storing the array "table[]". Algorithm 5 offers middle ground between the simple randomization algorithms (Algorithm 1 and Algorithm 2) and the hash-based algorithms (Algorithm 3 and Algorithm 4). The upper limit on the random increments (the value "N" in the pseudo-code included in Section 3.3.5) controls the trade-off between randomization and port- reuse frequency. Finally, a special case that may preclude the utilization of Algorithm 3 and Algorithm 4 should be analyzed. There exist some applications that contain the following code sequence: s = socket(); bind(s, IP_address, port = *); In some BSD-derived systems, the call to bind() will result in the selection of an ephemeral port number. However, as neither the remote IP address nor the remote port will be available to the ephemeral port selection function, the hash function F() used in Algorithm 3 and Algorithm 4 will not have all the required arguments, and thus the result of the hash function will be impossible to compute. Transport protocols implementing Algorithm 3 or Algorithm 4 should consider using Algorithm 2 when facing the scenario just described. An alternative to this behavior would be to implement "lazy binding" in response to the bind() call. That is, selection of an ephemeral port would be delayed until, e.g., connect() or send() are called. Thus, at that point the ephemeral port is actually selected, all the necessary arguments for the hash function F() are available, and therefore Algorithm 3 and Algorithm 4 could still be used in this scenario. This algorithm has been implemented by Linux [Linux].
4. Interaction with Network Address Port Translation (NAPT)
Network Address Port Translation (NAPT) translates both the network address and transport-protocol port number, thus allowing the transport identifiers of a number of private hosts to be multiplexed into the transport identifiers of a single external address [RFC2663]. In those scenarios in which a NAPT is present between the two endpoints of a transport-protocol instance, the obfuscation of the ephemeral port selection (from the point of view of the external network) will depend on the ephemeral port selection function at the NAPT. Therefore, NAPTs should consider obfuscating the selection of ephemeral ports by means of any of the algorithms discussed in this document. A NAPT that does not implement port preservation [RFC4787] [RFC5382] SHOULD obfuscate selection of the ephemeral port of a packet when it is changed during translation of that packet. A NAPT that does implement port preservation SHOULD obfuscate the ephemeral port of a packet only if the port must be changed as a result of the port being already in use for some other session. A NAPT that performs parity preservation and that must change the ephemeral port during translation of a packet SHOULD obfuscate the ephemeral ports. The algorithms described in this document could be easily adapted such that the parity is preserved (i.e., force the lowest order bit of the resulting port number to 0 or 1 according to whether even or odd parity is desired). Some applications allocate contiguous ports and expect to see contiguous ports in use at their peers. Clearly, this expectation might be difficult to accommodate at a NAPT, since some port numbers might already be in use by other sessions, and thus an alternative port might need to be selected, thus resulting in a non-contiguous port number sequence (see Section 4.2.3 of [RFC4787]). A NAPT that implements a simple port randomization algorithm (such as Algorithm 1, Algorithm 2, or Algorithm 5) is likely to break this assumption, even if the endpoint selecting an ephemeral port does select ephemeral ports that are contiguous. However, since a number of different ephemeral port selection algorithms have been implemented by deployed NAPTs, any application that relies on any specific ephemeral port selection algorithm at the NAPT is likely to suffer interoperability problems when a NAPT is present between the two endpoints of a transport-protocol instance. Nevertheless, some of the algorithms described in this document (namely Algorithm 3 and Algorithm 4) select consecutive ephemeral ports such that they are
contiguous (except when one of the port numbers needed to produce a contiguous sequence is already in use by some other NAPT session). Therefore, a NAPT willing to produce sequences of contiguous port numbers should consider implementing Algorithm 3 or Algorithm 4 of this document. Section 3.5 provides further guidance in choosing a port selection algorithm. It should be noted that in some network scenarios, a NAPT may naturally obscure ephemeral port selections simply due to the vast range of services with which it establishes connections and to the overall rate of the traffic [Allman].5. Security Considerations
Obfuscating the ephemeral port selection is no replacement for cryptographic mechanisms, such as IPsec [RFC4301], in terms of protecting transport-protocol instances against blind attacks. An eavesdropper that can monitor the packets that correspond to the transport-protocol instance to be attacked could learn the IP addresses and port numbers in use (and also sequence numbers, etc.) and easily perform an attack. Obfuscation of the ephemeral port selection does not provide any additional protection against this kind of attack. In such situations, proper authentication mechanisms such as those described in [RFC4301] should be used. This specification recommends including the whole range 1024-65535 for the selection of ephemeral ports, and suggests that an implementation maintains a list of those port numbers that should not be made available for ephemeral port selection. If the list of port numbers that are not available is significant, Algorithm 1 may be highly biased and generate predictable ports, as noted in Section 3.3.1. In particular, if the list of IANA Registered Ports is accepted as the local list of port numbers that should not be made available, certain ports may result with 500 times the probability of other ports. Systems that support numerous applications resulting in large lists of unavailable ports, or that use the IANA Registered Ports without modification, MUST NOT use Algorithm 1. If the local offset function F() (in Algorithm 3 and Algorithm 4) results in identical offsets for different inputs at greater frequency than would be expected by chance, the port-offset mechanism proposed in this document would have a reduced effect. If random numbers are used as the only source of the secret key, they should be chosen in accordance with the recommendations given in [RFC4086].
If an attacker uses dynamically assigned IP addresses, the current ephemeral port offset (Algorithm 3 and Algorithm 4) for a given five- tuple can be sampled and subsequently used to attack an innocent peer reusing this address. However, this is only possible until a re- keying happens as described above. Also, since ephemeral ports are only used on the client side (e.g., the one initiating the transport- protocol communication), both the attacker and the new peer need to act as servers in the scenario just described. While servers using dynamic IP addresses exist, they are not very common, and with an appropriate re-keying mechanism the effect of this attack is limited.6. Acknowledgements
The offset function used in Algorithm 3 and Algorithm 4 was inspired by the mechanism proposed by Steven Bellovin in [RFC1948] for defending against TCP sequence number attacks. The authors would like to thank (in alphabetical order) Mark Allman, Jari Arkko, Matthias Bethke, Stephane Bortzmeyer, Brian Carpenter, Vincent Deffontaines, Ralph Droms, Lars Eggert, Pasi Eronen, Gorry Fairhurst, Adrian Farrel, Guillermo Gont, David Harrington, Alfred Hoenes, Avshalom Houri, Charlie Kaufman, Amit Klein, Subramanian Moonesamy, Carlos Pignataro, Tim Polk, Kacheong Poon, Pasi Sarolahti, Robert Sparks, Randall Stewart, Joe Touch, Michael Tuexen, Magnus Westerlund, and Dan Wing for their valuable feedback on draft versions of this document. The authors would like to thank Alfred Hoenes for his admirable effort in improving the quality of this document. The authors would like to thank FreeBSD's Mike Silbersack for a very fruitful discussion about ephemeral port selection techniques. Fernando Gont's attendance to IETF meetings was supported by ISOC's "Fellowship to the IETF" program.7. References
7.1. Normative References
[RFC0768] Postel, J., "User Datagram Protocol", STD 6, RFC 768, August 1980. [RFC0793] Postel, J., "Transmission Control Protocol", STD 7, RFC 793, September 1981. [RFC1321] Rivest, R., "The MD5 Message-Digest Algorithm", RFC 1321, April 1992.
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", BCP 14, RFC 2119, March 1997. [RFC2385] Heffernan, A., "Protection of BGP Sessions via the TCP MD5 Signature Option", RFC 2385, August 1998. [RFC3550] Schulzrinne, H., Casner, S., Frederick, R., and V. Jacobson, "RTP: A Transport Protocol for Real-Time Applications", STD 64, RFC 3550, July 2003. [RFC3605] Huitema, C., "Real Time Control Protocol (RTCP) attribute in Session Description Protocol (SDP)", RFC 3605, October 2003. [RFC3828] Larzon, L-A., Degermark, M., Pink, S., Jonsson, L-E., and G. Fairhurst, "The Lightweight User Datagram Protocol (UDP-Lite)", RFC 3828, July 2004. [RFC4086] Eastlake, D., Schiller, J., and S. Crocker, "Randomness Requirements for Security", BCP 106, RFC 4086, June 2005. [RFC4301] Kent, S. and K. Seo, "Security Architecture for the Internet Protocol", RFC 4301, December 2005. [RFC4340] Kohler, E., Handley, M., and S. Floyd, "Datagram Congestion Control Protocol (DCCP)", RFC 4340, March 2006. [RFC4787] Audet, F. and C. Jennings, "Network Address Translation (NAT) Behavioral Requirements for Unicast UDP", BCP 127, RFC 4787, January 2007. [RFC4960] Stewart, R., "Stream Control Transmission Protocol", RFC 4960, September 2007. [RFC5382] Guha, S., Biswas, K., Ford, B., Sivakumar, S., and P. Srisuresh, "NAT Behavioral Requirements for TCP", BCP 142, RFC 5382, October 2008.7.2. Informative References
[Allman] Allman, M., "Comments On Selecting Ephemeral Ports", ACM Computer Communication Review, 39(2), 2009.
[CPNI-TCP] Gont, F., "CPNI Technical Note 3/2009: Security Assessment of the Transmission Control Protocol (TCP)", 2009, <http://www.cpni.gov.uk/Docs/ tn-03-09-security-assessment-TCP.pdf>. [FreeBSD] The FreeBSD Project, <http://www.freebsd.org>. [IANA] "IANA Port Numbers", <http://www.iana.org/assignments/port-numbers>. [Linux] The Linux Project, <http://www.kernel.org>. [NetBSD] The NetBSD Project, <http://www.netbsd.org>. [OpenBSD] The OpenBSD Project, <http://www.openbsd.org>. [OpenSolaris] OpenSolaris, <http://www.opensolaris.org>. [RFC1337] Braden, B., "TIME-WAIT Assassination Hazards in TCP", RFC 1337, May 1992. [RFC1948] Bellovin, S., "Defending Against Sequence Number Attacks", RFC 1948, May 1996. [RFC2663] Srisuresh, P. and M. Holdrege, "IP Network Address Translator (NAT) Terminology and Considerations", RFC 2663, August 1999. [RFC4953] Touch, J., "Defending TCP Against Spoofing Attacks", RFC 4953, July 2007. [RFC5925] Touch, J., Mankin, A., and R. Bonica, "The TCP Authentication Option", RFC 5925, June 2010. [RFC5927] Gont, F., "ICMP Attacks against TCP", RFC 5927, July 2010. [SCTP-SOCKET] Stewart, R., Poon, K., Tuexen, M., Lei, P., and V. Yasevich, V., "Sockets API Extensions for Stream Control Transmission Protocol (SCTP)", Work in Progress, January 2011. [Silbersack] Silbersack, M., "Improving TCP/IP security through randomization without sacrificing interoperability", EuroBSDCon 2005 Conference. [Stevens] Stevens, W., "Unix Network Programming, Volume 1: Networking APIs: Socket and XTI", Prentice Hall, 1998.
Appendix A. Survey of the Algorithms in Use by Some Popular Implementations
A.1. FreeBSD
FreeBSD 8.0 implements Algorithm 1, and in response to this document now uses a "min_port" of 10000 and a "max_port" of 65535 [FreeBSD].A.2. Linux
Linux 2.6.15-53-386 implements Algorithm 3, with MD5 as the hash algorithm. If the algorithm is faced with the corner-case scenario described in Section 3.5, Algorithm 1 is used instead [Linux].A.3. NetBSD
NetBSD 5.0.1 does not obfuscate its ephemeral port numbers. It selects ephemeral port numbers from the range 49152-65535, starting from port 65535, and decreasing the port number for each ephemeral port number selected [NetBSD].A.4. OpenBSD
OpenBSD 4.2 implements Algorithm 1, with a "min_port" of 1024 and a "max_port" of 49151. [OpenBSD]A.5. OpenSolaris
OpenSolaris 2009.06 implements Algorithm 1, with a "min_port" of 32768 and a "max_port" of 65535 [OpenSolaris].
Authors' Addresses
Michael Vittrup Larsen Tieto Skanderborgvej 232 Aarhus DK-8260 Denmark Phone: +45 8938 5100 EMail: michael.larsen@tieto.com Fernando Gont Universidad Tecnologica Nacional / Facultad Regional Haedo Evaristo Carriego 2644 Haedo, Provincia de Buenos Aires 1706 Argentina Phone: +54 11 4650 8472 EMail: fernando@gont.com.ar