rfc6024
Internet Engineering Task Force (IETF) R. Reddy
Request for Comments: 6024 National Security Agency
Category: Informational C. Wallace
ISSN: 2070-1721 Cygnacom Solutions
October 2010
Trust Anchor Management Requirements
Abstract
A trust anchor represents an authoritative entity via a public key
and associated data. The public key is used to verify digital
signatures, and the associated data is used to constrain the types of
information for which the trust anchor is authoritative. A relying
party uses trust anchors to determine if a digitally signed object is
valid by verifying a digital signature using the trust anchor's
public key, and by enforcing the constraints expressed in the
associated data for the trust anchor. This document describes some
of the problems associated with the lack of a standard trust anchor
management mechanism and defines requirements for data formats and
push-based protocols designed to address these problems.
Status of This Memo
This document is not an Internet Standards Track specification; it is
published for informational purposes.
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). Not all documents
approved by the IESG are a candidate for any level of Internet
Standard; see 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/rfc6024.
Copyright Notice
Copyright (c) 2010 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
Reddy & Wallace Informational [Page 1]
RFC 6024 Trust Anchor Management October 2010
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 . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1. Terminology . . . . . . . . . . . . . . . . . . . . . . . 4
1.2. Requirements Notation . . . . . . . . . . . . . . . . . . 4
2. Problem Statement . . . . . . . . . . . . . . . . . . . . . . 5
3. Requirements . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1. Transport Independence . . . . . . . . . . . . . . . . . . 6
3.2. Basic Management Operations . . . . . . . . . . . . . . . 7
3.3. Management Targets . . . . . . . . . . . . . . . . . . . . 7
3.4. Delegation of TA Manager Authority . . . . . . . . . . . . 8
3.5. RFC 5280 Support . . . . . . . . . . . . . . . . . . . . . 9
3.6. Support Purposes other than Certification Path
Validation . . . . . . . . . . . . . . . . . . . . . . . . 9
3.7. Trust Anchor Format . . . . . . . . . . . . . . . . . . . 10
3.8. Source Authentication . . . . . . . . . . . . . . . . . . 10
3.9. Reduce Reliance on Out-of-Band Trust Mechanisms . . . . . 11
3.10. Replay Detection . . . . . . . . . . . . . . . . . . . . . 11
3.11. Compromise or Disaster Recovery . . . . . . . . . . . . . 12
4. Security Considerations . . . . . . . . . . . . . . . . . . . 12
5. References . . . . . . . . . . . . . . . . . . . . . . . . . . 13
5.1. Normative References . . . . . . . . . . . . . . . . . . . 13
5.2. Informative References . . . . . . . . . . . . . . . . . . 13
Reddy & Wallace Informational [Page 2]
RFC 6024 Trust Anchor Management October 2010
1. Introduction
Digital signatures are used in many applications. For digital
signatures to provide integrity and authentication, the public key
used to verify the digital signature must be "trusted", i.e.,
accepted by a relying party (RP) as appropriate for use in the given
context. A public key used to verify a signature must be configured
as a trust anchor (TA) or contained in a certificate that can be
transitively verified by a certification path terminating at a trust
anchor. A trust anchor is a public key and associated data used by a
relying party to validate a signature on a signed object where the
object is either:
o a public key certificate that begins a certification path
terminated by a signature certificate or encryption certificate
o an object, other than a public key certificate or certificate
revocation list (CRL), that cannot be validated via use of a
certification path
Trust anchors have only local significance, i.e., each RP is
configured with a set of trust anchors, either by the RP or by an
entity that manages TAs in the context in which the RP operates. The
associated data defines the scope of a trust anchor by imposing
constraints on the signatures that the trust anchor may be used to
verify. For example, if a trust anchor is used to verify signatures
on X.509 certificates, these constraints may include a combination of
name spaces, certificate policies, or application/usage types.
One use of digital signatures is the verification of signatures on
firmware packages loaded into hardware modules, such as cryptographic
modules, cable boxes, routers, etc. Since such devices are often
managed remotely, the devices must be able to authenticate the source
of management interactions and can use trust anchors to perform this
authentication. However, trust anchors require management as well.
Other applications requiring trust anchor management include web
browsers (which use trust anchors when authenticating web servers)
and email clients (which use trust anchors when validating signed
email and when authenticating recipients of encrypted email).
All applications that rely upon digital signatures rely upon some
means of managing one or more sets of trust anchors. Each set of
trust anchors is referred to in this document as a trust anchor
store. Often, the means of managing trust anchor stores are
application-specific and rely upon out-of-band means to establish and
maintain trustworthiness. An application may use multiple trust
Reddy & Wallace Informational [Page 3]
RFC 6024 Trust Anchor Management October 2010
anchor stores, and a given trust anchor store may be used by multiple
applications. Each trust anchor store is managed by at least one TA
manager; a TA manager may manage multiple TA stores.
The requirements stated in this document were prepared prior to the
publication of [RFC5914] and [RFC5934]. The document was not
published at that time to allow for changes in requirements during
the development of the associated technical specifications. The
requirements described below are those that were considered during
the development of [RFC5914] and [RFC5934].
This section provides an introduction and defines basic terminology.
Section 2 describes problems with current trust anchor management
methods. Sections 3 and 4 describe requirements and security
considerations for a trust anchor management solution.
1.1. Terminology
The following terms are defined in order to provide a vocabulary for
describing requirements for trust anchor management.
Trust Anchor: A trust anchor represents an authoritative entity via
a public key and associated data. The public key is used to
verify digital signatures, and the associated data is used to
constrain the types of information for which the trust anchor is
authoritative. A relying party uses trust anchors to determine if
a digitally signed object is valid by verifying a digital
signature using the trust anchor's public key, and by enforcing
the constraints expressed in the associated data for the trust
anchor.
Trust Anchor Manager: A trust anchor manager is an entity
responsible for managing the contents of a trust anchor store.
Throughout this document, each trust anchor manager is assumed to
be represented as or delegated by a distinct trust anchor.
Trust Anchor Store: A trust anchor store is a set of one or more
trust anchors stored in a device. A trust anchor store may be
managed by one or more trust anchor managers. A device may have
more than one trust anchor store, each of which may be used by one
or more applications.
1.2. Requirements Notation
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].
Reddy & Wallace Informational [Page 4]
RFC 6024 Trust Anchor Management October 2010
2. Problem Statement
Trust anchors are used to support many application scenarios. Most
Internet browsers and email clients use trust anchors when
authenticating Transport Layer Security (TLS) sessions, verifying
signed email, and generating encrypted email by validating a
certification path to a server's certificate, an email originator's
certificate, or an email recipient's certificate, respectively. Many
software distributions are digitally signed to enable authentication
of the software source prior to installation. Trust anchors that
support these applications are typically installed as part of the
operating system (OS) or application, installed using an enterprise
configuration management system, or installed directly by an OS or
application user.
Trust anchors are typically stored in application-specific or
OS-specific trust anchor stores. Often, a single machine may have a
number of different trust anchor stores that may not be synchronized.
Reviewing the contents of a particular trust anchor store typically
involves use of a proprietary tool that interacts with a particular
type of trust store.
The presence of a trust anchor in a particular store often conveys
implicit authorization to validate signatures for any contexts from
which the store is accessed. For example, the public key of a
timestamp authority (TSA) may be installed in a trust anchor store to
validate signatures on timestamps [RFC3161]. However, if the store
containing this TA is used by multiple applications that serve
different purposes, the same key may be used (inappropriately) to
validate other types of objects such as certificates or Online
Certificate Status Protocol (OCSP) responses. Prior to publication
of [RFC5914], there was no standard general-purpose mechanism for
limiting the applicability (scope) of a trust anchor. A common
practice to address this problem is to place different TAs in
different stores and limit the set of applications that access a
given TA store.
Trust relationships between Public Key Infrastructures (PKIs) are
negotiated by policy authorities. Negotiations frequently require
significant time to ensure all participating parties' requirements
are satisfied. These requirements are expressed, to some extent, in
public key certificates via policy constraints, name constraints,
etc. In order for these requirements to be enforced, trust anchor
stores must be managed in accord with policy authority intentions.
Otherwise, the constraints defined in a cross-certificate could be
circumvented by recognizing the subject of the cross certificate as a
trust anchor, which would enable path processing implementations to
avoid the cross-certificate.
Reddy & Wallace Informational [Page 5]
RFC 6024 Trust Anchor Management October 2010
Trust anchors are often represented as self-signed certificates,
which provide no useful means of establishing the validity of the
information contained in the certificate. Confidence in the
integrity of a trust anchor is typically established through out-of-
band means, often by checking the "fingerprint" (one-way hash) of the
self-signed certificate with an authoritative source. Routine trust
anchor rekey operations typically require similar out-of-band checks,
though in-band rekey of a trust anchor is supported by the
Certificate Management Protocol (CMP) [RFC4210]. Ideally, only the
initial set of trust anchors are installed in a particular trust
anchor store should require out-of-band verification, particularly
when the costs of performing out-of-band checks commensurate with the
security requirements of applications using the trust anchor store
are high.
Despite the prevalent use of trust anchors, there is neither a
standard means for discovering the set of trust anchors installed in
a particular trust anchor store nor a standard means of managing
those trust anchors. The remainder of this document describes
requirements for a solution to this problem along with some security
considerations.
3. Requirements
This section describes the requirements for a trust anchor management
protocol. Requirements are provided for trust anchor contents as
well as for trust anchor store management operations.
3.1. Transport Independence
3.1.1. Functional Requirements
A general-purpose solution for the management of trust anchors MUST
be transport independent in order to apply to a range of device
communications environments. It MUST work in both session-oriented
and store-and-forward communications environments as well as in both
push and pull distribution models. To accommodate both communication
models in a uniform fashion, connectionless integrity and data origin
authentication for TA transactions MUST be provided at the
application layer. Confidentiality MAY be provided for such
transactions.
3.1.2. Rationale
Not all devices that use trust anchors are available for online
management operations; some devices may require manual interaction
for trust anchor management. Data origin authentication and
integrity are required to ensure that the transaction has not been
Reddy & Wallace Informational [Page 6]
RFC 6024 Trust Anchor Management October 2010
modified en route. Only connectionless integrity is required, for
compatibility with store-and-forward contexts.
3.2. Basic Management Operations
3.2.1. Functional Requirements
At a minimum, a protocol used for trust anchor management MUST enable
a trust anchor manager to perform the following operations:
o Determine which trust anchors are installed in a particular trust
anchor store
o Add one or more trust anchors to a trust anchor store
o Remove one or more trust anchors from a trust anchor store
o Replace an entire trust anchor store
A trust anchor management protocol MUST provide support for these
basic operations; however, not all implementations must support each
option. For example, some implementations may support only
replacement of trust anchor stores.
3.2.2. Rationale
These requirements describe the core operations required to manage
the contents of a trust anchor store. An edit operation was omitted
for the sake of simplicity, with consecutive remove and add
operations used for this purpose. A single add or remove operation
can act upon more than one trust anchor to avoid unnecessary round
trips and are provided to avoid the need to always replace an entire
trust anchor store. Trust anchor store replacement may be useful as
a simple, higher-bandwidth alternative to add and remove operations.
3.3. Management Targets
3.3.1. Functional Requirements
A protocol for TA management MUST allow a TA management transaction
to be directed to:
All TA stores for which the manager is responsible
An enumerated list of one or more named groups of trust anchor
stores
An individual trust anchor store
Reddy & Wallace Informational [Page 7]
RFC 6024 Trust Anchor Management October 2010
3.3.2. Rationale
Connections between PKIs can be accomplished using different means.
Unilateral or bilateral cross-certification can be performed, or a
community may simply elect to explicitly accept a trust anchor from
another community. Typically, these decisions occur at the
enterprise level. In some scenarios, it can be useful to establish
these connections for a small community within an enterprise.
Enterprise-wide mechanisms such as cross-certificates are ill-suited
for this purpose since certificate revocation or expiration affects
the entire enterprise.
A trust anchor management protocol can address this issue by
supporting limited installation of trust anchors (i.e., installation
of TAs in subsets of the enterprise user community), and by
supporting expression of constraints on trust anchor use by relying
parties. Limited installation requires the ability to identify the
members of the community that are intended to rely upon a particular
trust anchor, as well as the ability to query and report on the
contents of trust anchor stores. Trust anchor constraints can be
used to represent the limitations that might otherwise be expressed
in a cross-certificate, and limited installation ensures the
recognition of the trust anchor does not necessarily encompass an
entire enterprise.
Trust anchor configurations may be uniform across an enterprise, or
they may be unique to a single application or small set of
applications. Many devices and some applications utilize multiple
trust anchor stores. By providing means of addressing a specific
store or collections of stores, a trust anchor management protocol
can enable efficient management of all stores under a trust anchor
manager's control.
3.4. Delegation of TA Manager Authority
3.4.1. Functional Requirements
A trust anchor management protocol MUST enable secure transfer of
control of a trust anchor store from one trust anchor manager to
another. It also SHOULD enable delegation for specific operations
without requiring delegation of the overall trust anchor management
capability itself.
3.4.2. Rationale
Trust anchor manager rekey is one type of transfer that must be
supported. In this case, the new key will be assigned the same
privileges as the old key.
Reddy & Wallace Informational [Page 8]
RFC 6024 Trust Anchor Management October 2010
Creation of trust anchors for specific purposes, such as firmware
signing, is another example of delegation. For example, a trust
anchor manager may delegate only the authority to sign firmware to an
entity, but disallow further delegation of that privilege, or the
trust anchor manager may allow its delegate to further delegate
firmware signing authority to other entities.
3.5. RFC 5280 Support
3.5.1. Functional Requirements
A trust anchor management protocol MUST enable management of trust
anchors that will be used to validate certification paths and CRLs in
accordance with [RFC5280] and [RFC5055]. A trust anchor format MUST
enable the representation of constraints that influence certification
path validation or otherwise establish the scope of usage of the
trust anchor public key. Examples of such constraints are name
constraints, certificate policies, and key usage.
3.5.2. Rationale
Certification path validation is one of the most common applications
of trust anchors. The rules for using trust anchors for path
validation are established in [RFC5280]. [RFC5055] describes the use
of trust anchors for delegated path validation. Trust anchors used
to validate certification paths are responsible for providing,
possibly through a delegate, the revocation status information of
certificates it issues; this is often accomplished by signing a CRL.
3.6. Support Purposes other than Certification Path Validation
3.6.1. Functional Requirements
A trust anchor management protocol MUST enable management of trust
anchors that can be used for purposes other than certification path
validation, including trust anchors that cannot be used for
certification path validation. It SHOULD be possible to authorize a
trust anchor to delegate authority (to other TAs or certificate
holders) and to prevent a trust anchor from delegating authority.
3.6.2. Rationale
Trust anchors are used to validate a variety of signed objects, not
just public key certificates and CRLs. For example, a trust anchor
may be used to verify firmware packages [RFC4108], OCSP responses
[RFC2560], Server-Based Certificate Validation Protocol (SCVP)
responses [RFC5055], or timestamps [RFC3161]. TAs that are
authorized for use with some or all of these other types of
Reddy & Wallace Informational [Page 9]
RFC 6024 Trust Anchor Management October 2010
operations may not be authorized to verify public key certificates or
CRLs. Thus, it is important to be able to impose constraints on the
ways in which a given TA is employed.
3.7. Trust Anchor Format
3.7.1. Functional Requirements
Minimally, a trust anchor management protocol MUST support management
of trust anchors represented as self-signed certificates and trust
anchors represented as a distinguished name, public key information,
and, optionally, associated data. The definition of a trust anchor
MUST include a public key, a public key algorithm, and, if necessary,
public key parameters. When the public key is used to validate
certification paths or CRLs, a distinguished name also MUST be
included per [RFC5280]. A trust anchor format SHOULD enable
specification of a public key identifier to enable other applications
of the trust anchor, for example, verification of data signed using
the Cryptographic Message Syntax (CMS) SignedData structure
[RFC5652]. A trust anchor format also SHOULD enable the
representation of constraints that can be applied to restrict the use
of a trust anchor.
3.7.2. Rationale
Prior to the publication of [RFC5914], there was no standardized
format for trust anchors. Self-signed X.509 certificates are
typically used, but [RFC5280] does not mandate a particular trust
anchor representation. It requires only that a trust anchor's public
key information and distinguished name be available during
certification path validation. CMS is widely used to protect a
variety of types of content using digital signatures, including
contents that may be verified directly using a trust anchor, such as
firmware packages [RFC4108]. Constraints may include a validity
period, constraints on certification path validation, etc.
3.8. Source Authentication
3.8.1. Functional Requirements
An entity receiving trust anchor management data MUST be able to
authenticate the identity of the party providing the information and
MUST be able to confirm the party is authorized to provide that trust
anchor information.
Reddy & Wallace Informational [Page 10]
RFC 6024 Trust Anchor Management October 2010
A trust anchor manager MUST be able to authenticate which trust
anchor store corresponds to a report listing the contents of the
trust anchor store and be able to confirm the contents of the report
have not been subsequently altered.
3.8.2. Rationale
Data origin authentication and integrity are required to support
remote management operations, even when TA management transactions
are effected via store-and-forward communications.
3.9. Reduce Reliance on Out-of-Band Trust Mechanisms
3.9.1. Functional Requirements
When performing add operations, a trust anchor management protocol
SHOULD enable TA integrity to be checked automatically by a relying
party without relying on out-of-band trust mechanisms.
3.9.2. Rationale
Traditionally, a trust anchor is distributed out-of-band with its
integrity checked manually prior to installation. Installation
typically is performed by anyone with sufficient administrative
privilege on the system receiving the trust anchor. Reliance on out-
of-band trust mechanisms is one problem with current trust anchor
management approaches, and reduction of the need to use out-of-band
trust mechanisms is a primary motivation for developing a trust
anchor management protocol. Ideally, out-of-band trust mechanisms
will be required only during trust anchor store initialization.
3.10. Replay Detection
3.10.1. Functional Requirements
A trust anchor management protocol MUST enable participants engaged
in a trust anchor management protocol exchange to detect replay
attacks. A replay detection mechanism that does not introduce a
requirement for a reliable source of time MUST be available.
Mechanisms that do require a reliable source of time MAY be
available.
3.10.2. Rationale
Detection of replays of trust anchor management transactions is
required to support remote management operations. Replay of old
trust anchor management transactions could result in the
Reddy & Wallace Informational [Page 11]
RFC 6024 Trust Anchor Management October 2010
reintroduction of compromised trust anchors to a trust anchor store,
potentially exposing applications to malicious signed objects or
certification paths.
Some devices that utilize trust anchors have no access to a reliable
source of time, so a replay detection mechanism that requires a
reliable time source is insufficient.
3.11. Compromise or Disaster Recovery
3.11.1. Functional Requirements
A trust anchor management protocol MUST enable recovery from the
compromise or loss of a trust anchor private key, including the
private key authorized to serve as a trust anchor manager, without
requiring re-initialization of the trust store.
3.11.2. Rationale
Compromise or loss of a private key corresponding to a trust anchor
can have significant negative consequences. Currently, in some
cases, re-initialization of all affected trust anchor stores is
required to recover from a lost or compromised trust anchor key. Due
to the costs associated with re-initialization, a trust anchor
management protocol should support recovery options that do not
require trust anchor store re-initialization.
4. Security Considerations
The public key used to authenticate a TA management transaction may
have been placed in the client as the result of an earlier TA
management transaction or during an initial bootstrap configuration
operation. In most scenarios, at least one public key authorized for
trust anchor management must be placed in each trust anchor store to
be managed during the initial configuration of the trust anchor
store. This public key may be transported and checked using out-of-
band means. In all scenarios, regardless of the authentication
mechanism, at least one trust anchor manager must be established for
each trust anchor store during the initial configuration of the trust
anchor store.
Compromise of a trust anchor's private key can result in many
security problems including issuance of bogus certificates or
installation of rogue trust anchors.
Usage of trust anchor-based constraints requires great care when
defining trust anchors. Errors on the part of a trust anchor manager
could result in denial of service or have serious security
Reddy & Wallace Informational [Page 12]
RFC 6024 Trust Anchor Management October 2010
consequences. For example, if a name constraint for a trust anchor
that serves as the root of a PKI includes a typo, denial of service
results for certificate holders and relying parties. If a trust
anchor manager inadvertently delegates all of its privileges and the
delegate subsequently removes the trust anchor manager from trust
anchor stores now under its control, recovery may require
re-initialization of all effected trust anchor stores.
RFC 5280 requires that certificate path validation be initialized
with a TA subject name and public key, but does not require
processing of other information, such as name constraints extensions.
Inclusion of constraints in trust anchors is optional. When
constraints are explicitly included by a trust anchor manager using a
trust anchor management protocol, there exists an expectation that
the certificate path validation algorithm will make use of the
constraints. Application owners must confirm the path processing
implementations support the processing of TA-based constraints, where
required.
Many of the security considerations from [RFC5280] are also
applicable to trust anchor management.
5. References
5.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC5055] Freeman, T., Housley, R., Malpani, A., Cooper, D., and W.
Polk, "Server-Based Certificate Validation Protocol
(SCVP)", RFC 5055, December 2007.
[RFC5280] Cooper, D., Santesson, S., Farrell, S., Boeyen, S.,
Housley, R., and W. Polk, "Internet X.509 Public Key
Infrastructure Certificate and Certificate Revocation List
(CRL) Profile", RFC 5280, May 2008.
5.2. Informative References
[RFC2560] Myers, M., Ankney, R., Malpani, A., Galperin, S., and C.
Adams, "X.509 Internet Public Key Infrastructure Online
Certificate Status Protocol - OCSP", RFC 2560, June 1999.
[RFC3161] Adams, C., Cain, P., Pinkas, D., and R. Zuccherato,
"Internet X.509 Public Key Infrastructure Time-Stamp
Protocol (TSP)", RFC 3161, August 2001.
Reddy & Wallace Informational [Page 13]
RFC 6024 Trust Anchor Management October 2010
[RFC4108] Housley, R., "Using Cryptographic Message Syntax (CMS) to
Protect Firmware Packages", RFC 4108, August 2005.
[RFC4210] Adams, C., Farrell, S., Kause, T., and T. Mononen,
"Internet X.509 Public Key Infrastructure Certificate
Management Protocol (CMP)", RFC 4210, September 2005.
[RFC5652] Housley, R., "Cryptographic Message Syntax (CMS)", STD 70,
RFC 5652, September 2009.
[RFC5914] Housley, R., Ashmore, S., and C. Wallace, "Trust Anchor
Format", RFC 5914, June 2010.
[RFC5934] Housley, R., Ashmore, S., and C. Wallace, "Trust Anchor
Management Protocol (TAMP)", RFC 5934, August 2010.
Authors' Addresses
Raksha Reddy
National Security Agency
Suite 6599
9800 Savage Road
Fort Meade, MD 20755
EMail: r.reddy@radium.ncsc.mil
Carl Wallace
Cygnacom Solutions
Suite 5400
7925 Jones Branch Drive
McLean, VA 22102
EMail: cwallace@cygnacom.com
Reddy & Wallace Informational [Page 14]
ERRATA
