This document is obsolete. Please
refer to RFC 9293.
Internet Engineering Task Force (IETF) F. Gont Request for Comments: 6093 UTN/FRH Updates: 793, 1011, 1122 A. Yourtchenko Category: Standards Track Cisco ISSN: 2070-1721 January 2011
On the Implementation of the TCP Urgent Mechanism
Abstract
This document analyzes how current TCP implementations process TCP urgent indications and how the behavior of some widely deployed middleboxes affects how end systems process urgent indications. This document updates the relevant specifications such that they accommodate current practice in processing TCP urgent indications, raises awareness about the reliability of TCP urgent indications in the Internet, and recommends against the use of urgent indications (but provides advice to applications that do).
Status of This Memo
This is an Internet Standards Track document.
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 Internet Standards 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/rfc6093.
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.
Gont & Yourtchenko Standards Track [Page 1]
RFC 6093 On the TCP Urgent Mechanism January 2011
Table of Contents
1. Introduction ....................................................3 2. Specification of the TCP Urgent Mechanism .......................3 2.1. Semantics of Urgent Indications ............................3 2.2. Semantics of the Urgent Pointer ............................4 2.3. Allowed Length of "Urgent Data" ............................4 3. Current Implementation Practice of the TCP Urgent Mechanism .....5 3.1. Semantics of Urgent Indications ............................5 3.2. Semantics of the Urgent Pointer ............................5 3.3. Allowed Length of "Urgent Data" ............................6 3.4. Interaction of Middleboxes with TCP Urgent Indications .....6 4. Updating RFC 793, RFC 1011, and RFC 1122 ........................6 5. Advice to New Applications Employing TCP ........................7 6. Advice to Applications That Make Use of the Urgent Mechanism ....7 7. Security Considerations .........................................7 8. Acknowledgements ................................................8 9. References ......................................................8 9.1. Normative References .......................................8 9.2. Informative References .....................................8 Appendix A. Survey of the Processing of TCP Urgent Indications by Some Popular TCP Implementations ......10 A.1. FreeBSD ...................................................10 A.2. Linux .....................................................10 A.3. NetBSD ....................................................10 A.4. OpenBSD ...................................................11 A.5. Cisco IOS software ........................................11 A.6. Microsoft Windows 2000, Service Pack 4 ....................11 A.7. Microsoft Windows 2008 ....................................11 A.8. Microsoft Windows 95 ......................................11
This document analyzes how some current TCP implementations process TCP urgent indications, and how the behavior of some widely deployed middleboxes affects the processing of urgent indications by hosts. This document updates RFC 793 [RFC0793], RFC 1011 [RFC1011], and RFC 1122 [RFC1122] such that they accommodate current practice in processing TCP urgent indications. It also provides advice to applications using the urgent mechanism and raises awareness about the reliability of TCP urgent indications in the current Internet.
Given the above issues and potential interoperability issues with respect to the currently common default mode operation, it is strongly recommended that applications do not employ urgent indications. Nevertheless, urgent indications are still retained as a mandatory part of the TCP protocol to support the few legacy applications that employ them. However, it is expected that even these applications will have difficulties in environments with middleboxes.
Section 2 describes what the current IETF specifications state with respect to TCP urgent indications. Section 3 describes how current TCP implementations actually process TCP urgent indications. Section 4 updates RFC 793 [RFC0793], RFC 1011 [RFC1011], and RFC 1122 [RFC1122], such that they accommodate current practice in processing TCP urgent indications. Section 5 provides advice to new applications employing TCP, with respect to the TCP urgent mechanism. Section 6 provides advice to existing applications that use or rely on the TCP urgent mechanism.
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].
TCP implements an "urgent mechanism" that allows the sending user to stimulate the receiving user to accept some "urgent data" and that permits the receiving TCP to indicate to the receiving user when all the currently known "urgent data" have been read.
The TCP urgent mechanism permits a point in the data stream to be designated as the end of urgent information. Whenever this point is in advance of the receive sequence number (RCV.NXT) at the receiving TCP, that TCP must tell the user to go into "urgent mode"; when the receive sequence number catches up to the urgent pointer, the TCP
Gont & Yourtchenko Standards Track [Page 3]
RFC 6093 On the TCP Urgent Mechanism January 2011
must tell user to go into "normal mode" [RFC0793]. This means, for example, that data that was received as "normal data" might become "urgent data" if an urgent indication is received in some successive TCP segment before that data is consumed by the TCP user.
The URG control flag indicates that the "Urgent Pointer" field is meaningful and must be added to the segment sequence number to yield the urgent pointer. The absence of this flag indicates that there is no "urgent data" outstanding [RFC0793].
The TCP urgent mechanism is NOT a mechanism for sending "out-of-band" data: the so-called "urgent data" should be delivered "in-line" to the TCP user.
There is some ambiguity in RFC 793 [RFC0793] with respect to the semantics of the Urgent Pointer. Section 3.1 (page 17) of RFC 793 [RFC0793] states that the Urgent Pointer "communicates the current value of the urgent pointer as a positive offset from the sequence number in this segment. The urgent pointer points to the sequence number of the octet following the urgent data. This field is only be interpreted in segments with the URG control bit set" (sic). However, Section 3.9 (page 56) of RFC 793 [RFC0793] states, when describing the processing of the SEND call in the ESTABLISHED and CLOSE-WAIT states, that "If the urgent flag is set, then SND.UP <- SND.NXT-1 and set the urgent pointer in the outgoing segments".
RFC 1011 [RFC1011] clarified this ambiguity in RFC 793 stating that "Page 17 is wrong. The urgent pointer points to the last octet of urgent data (not to the first octet of non-urgent data)". RFC 1122 [RFC1122] formally updated RFC 793 by stating, in Section 4.2.2.4 (page 84), that "the urgent pointer points to the sequence number of the LAST octet (not LAST+1) in a sequence of urgent data".
RFC 793 [RFC0793] allows TCP peers to send "urgent data" of any length, as the TCP urgent mechanism simply provides a pointer to an interesting point in the data stream. In this respect, Section 4.2.2.4 (page 84) of RFC 1122 [RFC1122] explicitly states that "A TCP MUST support a sequence of urgent data of any length".
Gont & Yourtchenko Standards Track [Page 4]
RFC 6093 On the TCP Urgent Mechanism January 2011
3. Current Implementation Practice of the TCP Urgent Mechanism
As discussed in Section 2, the TCP urgent mechanism simply permits a point in the data stream to be designated as the end of urgent information but does NOT provide a mechanism for sending "out-of- band" data.
Unfortunately, virtually all TCP implementations process TCP urgent indications differently. By default, the last byte of "urgent data" is delivered "out of band" to the application. That is, it is not delivered as part of the normal data stream [UNPv1]. For example, the "out-of-band" byte is read by an application when a recv(2) system call with the MSG_OOB flag set is issued.
Most implementations provide a socket option (SO_OOBINLINE) that allows an application to override the (broken) default processing of urgent indications, so that "urgent data" is delivered "in line" to the application, thus providing the semantics intended by the IETF specifications.
All the popular implementations that the authors of this document have been able to test interpret the semantics of the TCP Urgent Pointer as specified in Section 3.1 of RFC 793. This means that even when RFC 1122 formally updated RFC 793 to clarify the ambiguity in the semantics of the Urgent Pointer, this clarification was never reflected in actual implementations (i.e., virtually all implementations default to the semantics of the urgent pointer specified in Section 3.1 of RFC 793).
Some operating systems provide a system-wide toggle to override this behavior and interpret the semantics of the Urgent Pointer as clarified in RFC 1122. However, this system-wide toggle has been found to be inconsistent. For example, Linux provides the sysctl "tcp_stdurg" (i.e., net.ipv4.tcp_stdurg) that, when set, supposedly changes the system behavior to interpret the semantics of the TCP Urgent Pointer as specified in RFC 1122. However, this sysctl changes the semantics of the Urgent Pointer only for incoming segments (i.e., not for outgoing segments). This means that if this sysctl is set, an application might be unable to interoperate with itself if both the TCP sender and the TCP receiver are running on the same host.
While Section 4.2.2.4 (page 84) of RFC 1122 explicitly states that "A TCP MUST support a sequence of urgent data of any length", in practice, all those implementations that interpret TCP urgent indications as a mechanism for sending "out-of-band" data keep a buffer of a single byte for storing the "last byte of urgent data". Thus, if successive indications of "urgent data" are received before the application reads the pending "out-of-band" byte, that pending byte will be discarded (i.e., overwritten by the new byte of "urgent data").
In order to avoid "urgent data" from being discarded, some implementations queue each of the received "urgent bytes", so that even if another urgent indication is received before the pending "urgent data" are consumed by the application, those bytes do not need to be discarded. Some of these implementations have been known to fail to enforce any limits on the amount of "urgent data" that they queue; thus, they become vulnerable to trivial resource exhaustion attacks [CPNI-TCP].
It should be reinforced that the aforementioned implementations are broken. The TCP urgent mechanism is not a mechanism for delivering "out-of-band" data.
3.4. Interaction of Middleboxes with TCP Urgent Indications
As a result of the publication of Network Intrusion Detection System (NIDS) evasion techniques based on TCP urgent indications [phrack], some middleboxes clear the urgent indications by clearing the URG flag and setting the Urgent Pointer to zero. This causes the "urgent data" to become "in line" (that is, accessible by the read(2) call or the recv(2) call without the MSG_OOB flag) in the case of those TCP implementations that interpret the TCP urgent mechanism as a facility for delivering "out-of-band" data (as described in Section 3.1). An example of such a middlebox is the Cisco PIX firewall [Cisco-PIX]. This should discourage applications from depending on urgent indications for their correct operation, as urgent indications may not be reliable in the current Internet.
Considering that as long as both the TCP sender and the TCP receiver implement the same semantics for the Urgent Pointer there is no functional difference in having the Urgent Pointer point to "the sequence number of the octet following the urgent data" vs. "the last octet of urgent data", and that all known implementations interpret the semantics of the Urgent Pointer as pointing to "the
Gont & Yourtchenko Standards Track [Page 6]
RFC 6093 On the TCP Urgent Mechanism January 2011
sequence number of the octet following the urgent data", we hereby update RFC 793 [RFC0793], RFC 1011 [RFC1011], and RFC 1122 [RFC1122] such that "the urgent pointer points to the sequence number of the octet following the urgent data" (in segments with the URG control bit set), thus accommodating virtually all existing TCP implementations.
As a result of the issues discussed in Section 3.2 and Section 3.4, new applications SHOULD NOT employ the TCP urgent mechanism. However, TCP implementations MUST still include support for the urgent mechanism such that existing applications can still use it.
6. Advice to Applications That Make Use of the Urgent Mechanism
Even though applications SHOULD NOT employ the urgent mechanism, applications that still decide to employ it MUST set the SO_OOBINLINE socket option, such that "urgent data" is delivered in line, as intended by the IETF specifications.
Additionally, applications that still decide to use the urgent mechanism need to be designed for correct operation even when the URG flag is cleared by middleboxes.
Multiple factors can affect the data flow that is actually delivered to an application when the TCP urgent mechanism is employed: for example, the two possible interpretations of the semantics of the Urgent Pointer in current implementations (e.g., depending on the value of the tcp_stdurg sysctl), the possible implementation of the urgent mechanism as an "out-of-band" (OOB) facility (versus "in-band" as intended by the IETF specifications), or middleboxes (such as packet scrubbers) or the end-systems themselves that could cause the "urgent data" to be processed "in line". This might make it difficult for a Network Intrusion Detection System (NIDS) to track the application-layer data transferred to the destination system and thus lead to false negatives or false positives in the NIDS [CPNI-TCP] [phrack].
Probably the best way to avoid the security implications of TCP "urgent data" is to avoid having applications use the TCP urgent mechanism altogether. Packet scrubbers could probably be configured to clear the URG bit and set the Urgent Pointer to zero. This would basically cause the "urgent data" to be put "in line". However, this
Gont & Yourtchenko Standards Track [Page 7]
RFC 6093 On the TCP Urgent Mechanism January 2011
might cause interoperability problems or undesired behavior in those applications that rely on the TCP urgent mechanism, such as Telnet [RFC0854] and FTP [RFC0959].
The authors of this document would like to thank (in alphabetical order) Jari Arkko, Ron Bonica, David Borman, Dave Cridland, Ralph Droms, Wesley Eddy, John Heffner, Alfred Hoenes, Alexey Melnikov, Keith Moore, Carlos Pignataro, Tim Polk, Anantha Ramaiah, Joe Touch, Michael Welzl, Dan Wing, and Alexander Zimmermann for providing valuable feedback on earlier versions of this document.
Fernando would like to thank David Borman and Joe Touch for a fruitful discussion about the TCP urgent mechanism at IETF 73 (Minneapolis).
Fernando Gont's attendance to IETF meetings was supported by ISOC's "Fellowship to the IETF" program.
Finally, Fernando Gont wishes to express deep and heartfelt gratitude to Jorge Oscar Gont and Nelida Garcia for their precious motivation and guidance.
[CPNI-TCP] Gont, F., "Security Assessment of the Transmission Control Protocol (TCP)", "http://www.cpni.gov.uk/ Docs/tn-03-09-security-assessment-TCP.pdf", 2009.
FreeBSD 8.0 [FreeBSD] interprets the semantics of the urgent pointer as specified in Section 4 of this document. It does not provide any sysctl to override this behavior.
FreeBSD provides the SO_OOBINLINE socket option that, when set, causes TCP "urgent data" to remain "in line". That is, it will be accessible by the read(2) call or the recv(2) call without the MSG_OOB flag.
FreeBSD supports only one byte of "urgent data". That is, only the byte preceding the Urgent Pointer is considered "urgent data".
Linux 2.6.15-53-386 [Linux] interprets the semantics of the urgent pointer as specified in Section 4 of this document. It provides the net.ipv4.tcp_stdurg sysctl to override this behavior to interpret the Urgent Pointer as specified in RFC 1122 [RFC1122]. However, this sysctl only affects the processing of incoming segments (the Urgent Pointer in outgoing segments will still be set as specified in Section 4 of this document).
Linux provides the SO_OOBINLINE socket option that, when set, causes TCP "urgent data" to remain "in line". That is, it will be accessible by the read(2) call or the recv(2) call without the MSG_OOB flag.
Linux supports only one byte of "urgent data". That is, only the byte preceding the Urgent Pointer is considered "urgent data".
NetBSD 5.0.1 [NetBSD] interprets the semantics of the urgent pointer as specified in Section 4 of this document. It does not provide any sysctl to override this behavior.
NetBSD provides the SO_OOBINLINE socket option that, when set, causes TCP "urgent data" to remain "in line". That is, it will be accessible by the read(2) call or the recv(2) call without the MSG_OOB flag.
NetBSD supports only one byte of "urgent data". That is, only the byte preceding the Urgent Pointer is considered "urgent data".
OpenBSD 4.2 [OpenBSD] interprets the semantics of the urgent pointer as specified in Section 4 of this document. It does not provide any sysctl to override this behavior.
OpenBSD provides the SO_OOBINLINE socket option that, when set, causes TCP "urgent data" to remain "in line". That is, it will be accessible by the read(2) or recv(2) calls without the MSG_OOB flag.
OpenBSD supports only one byte of "urgent data". That is, only the byte preceding the Urgent Pointer is considered "urgent data".
Microsoft Windows 2000 [Windows2000] interprets the semantics of the urgent pointer as specified in Section 4 of this document. It provides the TcpUseRFC1122UrgentPointer system-wide variable to override this behavior, interpreting the Urgent Pointer as specified in RFC 1122 [RFC1122].
Tests performed with a sample server application compiled using the cygwin environment has shown that the default behavior is to return the "urgent data" "in line".
Microsoft Windows 95 interprets the semantics of the urgent pointer as specified in Section 4 of this document. It provides the BSDUrgent system-wide variable to override this behavior, interpreting the Urgent Pointer as specified in RFC 1122 [RFC1122]. Windows 95 supports only one byte of "urgent data". That is, only the byte preceding the Urgent Pointer is considered "urgent data" [Windows95].
Gont & Yourtchenko Standards Track [Page 11]
RFC 6093 On the TCP Urgent Mechanism January 2011
Authors' Addresses
Fernando Gont Universidad Tecnologica Nacional / Facultad Regional Haedo Evaristo Carriego 2644 Haedo, Provincia de Buenos Aires 1706 Argentina