Internet-Draft | Service Identity | June 2023 |
Saint-Andre & Salz | Expires 21 December 2023 | [Page] |
Many application technologies enable secure communication between two entities by means of Transport Layer Security (TLS) with Internet Public Key Infrastructure Using X.509 (PKIX) certificates. This document specifies procedures for representing and verifying the identity of application services in such interactions.¶
This document obsoletes RFC 6125.¶
This note is to be removed before publishing as an RFC.¶
Discussion of this document takes place on the Using TLS in Applications Working Group mailing list (uta@ietf.org), which is archived at https://mailarchive.ietf.org/arch/browse/uta/.¶
Source for this draft and an issue tracker can be found at https://github.com/richsalz/draft-ietf-uta-rfc6125bis.¶
This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.¶
Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet-Drafts is at https://datatracker.ietf.org/drafts/current/.¶
Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."¶
This Internet-Draft will expire on 21 December 2023.¶
Copyright (c) 2023 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 (https://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 Revised BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Revised BSD License.¶
The visible face of the Internet largely consists of services that employ a client-server architecture in which a client communicates with an application service. When a client communicates with an application service using [TLS], [DTLS], or a protocol built on those ([QUIC] being a notable example), it has some notion of the server's identity (e.g., "the website at example.com") while attempting to establish secure communication. Likewise, during TLS negotiation, the server presents its notion of the service's identity in the form of a public-key certificate that was issued by a certificate authority (CA) in the context of the Internet Public Key Infrastructure using X.509 [PKIX]. Informally, we can think of these identities as the client's "reference identity" and the server's "presented identity"; more formal definitions are given later. A client needs to verify that the server's presented identity matches its reference identity so it can deterministically and automatically authenticate the communication.¶
This document defines procedures for how clients do this verification. It therefore also defines requirements on other parties, such as the certificate authorities that issue certificates, the service administrators requesting them, and the protocol designers defining how things are named.¶
This document obsoletes RFC 6125. Changes from RFC 6125 are described under Appendix A.¶
This document does not supersede the rules for certificate issuance or validation specified by [PKIX]. That document also governs any certificate-related topic on which this document is silent. This includes certificate syntax, extensions such as name constraints or extended key usage, and handling of certification paths.¶
This document addresses only name forms in the leaf "end entity" server certificate. It does not address the name forms in the chain of certificates used to validate a certificate, let alone creating or checking the validity of such a chain. In order to ensure proper authentication, applications need to verify the entire certification path.¶
The previous version of this specification, [VERIFY], surveyed the then-current practice from many IETF standards and tried to generalize best practices (see Appendix A of [VERIFY] for details).¶
This document takes the lessons learned since then and codifies them. The following is a summary of the rules, which are described at greater length in the remainder of this document:¶
This document applies only to service identities that are used in TLS or DTLS and that are included in PKIX certificates.¶
With regard to TLS and DTLS, these security protocols are used to protect data exchanged over a wide variety of application protocols, which use both the TLS or DTLS handshake protocol and the TLS or DTLS record layer, either directly or through a profile as in Network Time Security [NTS]. The TLS handshake protocol can also be used with different record layers to define secure transport protocols; at present the most prominent example is QUIC [RFC9000]. The rules specified here are intended to apply to all protocols in this extended TLS "family".¶
With regard to PKIX certificates, the primary usage is in the context of the public key infrastructure described in [PKIX]. In addition, technologies such as DNS-Based Authentication of Named Entities (DANE) [DANE] sometimes use certificates based on PKIX (more precisely, certificates structured via [X.509] or specific encodings thereof such as [X.690]), at least in certain modes. Alternatively, a TLS peer could issue delegated credentials that are based on a CA-issued certificate, as in [TLS-SUBCERTS]. In both cases, a TLS client could learn of a service identity through its inclusion in the relevant certificate. The rules specified here are intended to apply whenever service identities are included in X.509 certificates or credentials that are derived from such certificates.¶
The following topics are out of scope for this specification:¶
Certification authority policies. This includes items such as the following:¶
Because many concepts related to "identity" are often too vague to be actionable in application protocols, we define a set of more concrete terms for use in this specification.¶
A service on the Internet that enables clients to connect for the purpose of retrieving or uploading information, communicating with other entities, or connecting to a broader network of services.¶
An entity that hosts or deploys an application service.¶
A formal identifier for the application protocol used to provide a particular kind of application service at a domain. This often appears as a URI scheme [URI], DNS SRV Service [DNS-SRV], or an ALPN [ALPN] identifier.¶
A domain name or host name that is explicitly configured at the application layer for communicating with the source domain (e.g., by the human user controlling an application client, by a trusted administrator who sets policy for such clients, or by means of automated configuration in the client itself). For example, an IMAP server at mail.example.net could be a delegated domain for a source domain of example.net associated with an email address of user@example.net. This kind of application-layer delegation is not to be confused with delegation in the DNS, by which a separate zone is created in the name space beneath the apex of a given domain; see for instance [DNS-TERMS].¶
A domain name or host name that a client has derived from the source domain in an automated fashion (e.g., by means of an MX or SRV lookup).¶
A particular instance of an identifier type that is either presented by a server in a certificate or referenced by a client for matching purposes.¶
A formally defined category of identifier that can be included in a certificate and therefore that can also be used for matching purposes. For conciseness and convenience, we define the following identifier types of interest:¶
The short name for the Internet Public Key Infrastructure using X.509 defined in [PKIX]. That document provides a profile of the X.509v3 certificate specifications and X.509v2 certificate revocation list (CRL) specifications for use on the Internet.¶
An identifier presented by a server to a client within a PKIX certificate when the client attempts to establish secure communication with the server. The certificate can include one or more presented identifiers of different types, and if the server hosts more than one domain then the certificate might present distinct identifiers for each domain.¶
An identifier used by the client when examining presented identifiers. It is constructed from the source domain, and optionally an application service type.¶
An ASN.1-based construction which itself is a building-block component of Distinguished Names. See [LDAP-DN], Section 2.¶
The fully qualified domain name (FQDN) that a client expects an application service to present in the certificate. This is typically input by a human user, configured into a client, or provided by reference such as a URL. The combination of a source domain and, optionally, an application service type enables a client to construct one or more reference identifiers. This specification covers FQDNs. Use of any names that are not fully qualified is out of scope and may result in unexpected or undefined behavior.¶
An identifier placed in a subjectAltName extension.¶
A standard PKIX extension enabling identifiers of various types to be bound to the certificate subject.¶
The name of a PKIX certificate's subject, encoded in a certificate's subject field (see [PKIX], Section 4.1.2.6).¶
TLS uses the words client and server, where the client is the entity that initiates the connection. In many cases, this is consistent with common practice, such as a browser connecting to a Web origin. For the sake of clarity, and to follow the usage in [TLS] and related specifications, we will continue to use the terms client and server in this document. However, these are TLS-layer roles, and the application protocol could support the TLS server making requests to the TLS client after the TLS handshake; there is no requirement that the roles at the application layer match the TLS layer.¶
Security-related terms used in this document, but not defined here or in [PKIX] should be understood in the sense defined in [SECTERMS]. Such terms include "attack", "authentication", "identity", "trust", "validate", and "verify".¶
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all capitals, as shown here.¶
This document assumes that an application service is identified by a DNS domain
name (e.g., example.com
), an IP address (IPv4 or IPv6), or by an identifier
that contains additional supplementary information. Supplementary information
is limited to the application service type as expressed in a DNS SRV record
(e.g., "the IMAP server at example.net" for "_imap.example.net") or a URI.¶
In a DNS-ID - and in the DNS domain name portion of an SRV-ID or URI-ID - any characters outside the [US-ASCII] range are prohibited and internationalized domain labels are represented as A-labels [IDNA-DEFS].¶
An IP address is either a 4-octet IPv4 address [IPv4] or a 16-octet IPv6 address [IPv6]. The identifier might need to be converted from a textual representation to obtain this value.¶
From the perspective of the application client or user, some identifiers are direct because they are provided directly by a human user. This includes runtime input, prior configuration, or explicit acceptance of a client communication attempt. Other names are indirect because they are automatically resolved by the application based on user input, such as a target name resolved from a source name using DNS SRV or [NAPTR] records. The distinction matters most for certificate consumption, specifically verification as discussed in this document.¶
From the perspective of the application service, some identifiers are unrestricted because they can be used in any type of service, such as a single certificate being used for both the HTTP and IMAP services at the host "example.com". Other identifiers are restricted because they can only be used for one type of service, such as a special-purpose certificate that can only be used for an IMAP service. This distinction matters most for certificate issuance.¶
We can categorize the four identifier types as follows:¶
It is important to keep these distinctions in mind, because best practices for the deployment and use of the identifiers differ. Note that cross-protocol attacks such as [ALPACA] are possible when two different protocol services use the same certificate. This can be addressed by using restricted identifiers or deploying services so that they do not share certificates. Protocol specifications MUST specify which identifiers are mandatory-to-implement and SHOULD provide operational guidance when necessary.¶
The Common Name RDN MUST NOT be used to identify a service because it is not strongly typed (essentially free-form text) and therefore suffers from ambiguities in interpretation.¶
For similar reasons, other RDNs within the subjectName MUST NOT be used to identify a service.¶
An IP address that is the result of a DNS query is not direct. Use of IP-IDs that are not direct is out of scope for this document.¶
This section defines how protocol designers should reference this document, which would typically be a normative reference in their specification. Its specification MAY choose to allow only one of the identifier types defined here.¶
If the technology does not use DNS SRV records to resolve the DNS domain names of application services, then its specification MUST state that SRV-ID as defined in this document is not supported. Note that many existing application technologies use DNS SRV records to resolve the DNS domain names of application services, but do not rely on representations of those records in PKIX certificates by means of SRV-IDs as defined in [SRVNAME].¶
If the technology does not use URIs to identify application services, then its specification MUST state that URI-ID as defined in this document is not supported. Note that many existing application technologies use URIs to identify application services, but do not rely on representation of those URIs in PKIX certificates by means of URI-IDs.¶
A technology MAY disallow the use of the wildcard character in presented identifiers. If it does so, then the specification MUST state that wildcard certificates as defined in this document are not supported.¶
A protocol can allow the use of an IP address in place of a DNS name. This might use the same field without distinguishing the type of identifier, as for example in the "host" components of a URI. In this case, applications need to be aware that the textual representation of an IPv4 address can is a valid DNS name. The two types can be distinguished by first testing if the identifier is a valid IPv4 address, as is done by the "first-match-wins" algorithm in Section 3.2.2 of [URI]. Note also that by policy, Top-Level Domains ([DNS-TERMS]) do not start with a digit (see Section 2.2.1.3.2 of [ICANN-AGB]); historically this rule was also intended to apply to all labels in a domain name (see Section 2.3.1 of [DNS-NAMES]), although that is not always the case in practice.¶
This section provides instructions for issuers of certificates.¶
When a certificate authority issues a certificate based on the FQDN at which the application service provider will provide the relevant application, the following rules apply to the representation of application service identities. Note that some of these rules are cumulative and can interact in important ways that are illustrated later in this document.¶
sip
but not sips
or tel
for SIP as described in [SIP-SIPS]). Typically this
identifier type would supplement the DNS-ID, unless the certificate
is meant to be scoped to only the protocol in question.¶
Consider a simple website at www.example.com
, which is not discoverable via
DNS SRV lookups. Because HTTP does not specify the use of URIs in server
certificates, a certificate for this service might include only a DNS-ID of
www.example.com
.¶
Consider the same website, which is reachable by a fixed IP address of
2001:db8::5c
. The certificate might include this value in an IP-ID to allow
clients to use the fixed IP address as a reference identity.¶
Consider an IMAP-accessible email server at the host mail.example.net
servicing email addresses of the form user@example.net
and discoverable via
DNS SRV lookups on the application service name of example.net
. A
certificate for this service might include SRV-IDs of _imap.example.net
and
_imaps.example.net
(see [EMAIL-SRV]) along with DNS-IDs of example.net
and mail.example.net
.¶
Consider a SIP-accessible voice-over-IP (VoIP) server at the host
voice.example.edu
servicing SIP addresses of the form
user@voice.example.edu
and identified by a URI of <sip:voice.example.edu>.
A certificate for this service would include a URI-ID of
sip:voice.example.edu
(see [SIP-CERTS]) along with a DNS-ID of
voice.example.edu
.¶
Consider an XMPP-compatible instant messaging (IM) server at the host
im.example.org
servicing IM addresses of the form user@im.example.org
and
discoverable via DNS SRV lookups on the im.example.org
domain. A
certificate for this service might include SRV-IDs of
_xmpp-client.im.example.org
and _xmpp-server.im.example.org
(see
[XMPP]), a DNS-ID of im.example.org
.¶
This section provides instructions for service providers regarding the information to include in certificate signing requests (CSRs). In general, service providers SHOULD request certificates that include all the identifier types that are required or recommended for the application service type that will be secured using the certificate to be issued.¶
A service provider SHOULD request certificates with as few identifiers as necessary to identify a single service; see Section 7.5.¶
If the certificate will be used for only a single type of application service, the service provider SHOULD request a certificate that includes DNS-ID or IP-ID values that identify that service or, if appropriate for the application service type, SRV-ID or URI-ID values that limit the deployment scope of the certificate to only the defined application service type.¶
If the certificate might be used for any type of application service, then the service provider SHOULD request a certificate that includes only DNS-IDs or IP-IDs. Again, because of multi-protocol attacks this practice is discouraged; this can be mitigated by deploying only one service on a host.¶
If a service provider offers multiple application service types and wishes to
limit the applicability of certificates using SRV-IDs or URI-IDs, they SHOULD
request multiple certificates, rather than a single certificate containing
multiple SRV-IDs or URI-IDs each identifying a different application service
type. This rule does not apply to application service type "bundles" that
identify distinct access methods to the same underlying application such as
an email application with access methods denoted by the application service
types of imap
, imaps
, pop3
, pop3s
, and submission
as described in
[EMAIL-SRV].¶
At a high level, the client verifies the application service's identity by performing the following actions:¶
Naturally, in addition to checking identifiers, a client should perform further checks, such as expiration and revocation, to ensure that the server is authorized to provide the requested service. Because such checking is not a matter of verifying the application service identity presented in a certificate, methods for doing so are out of scope for this document.¶
The client MUST construct a list of acceptable reference identifiers, and MUST do so independently of the identifiers presented by the service.¶
The inputs used by the client to construct its list of reference identifiers might be a URI that a user has typed into an interface (e.g., an HTTPS URL for a website), configured account information (e.g., the domain name of a host for retrieving email, which might be different from the DNS domain name portion of a username), a hyperlink in a web page that triggers a browser to retrieve a media object or script, or some other combination of information that can yield a source domain and an application service type.¶
This document does not precisely define how reference identifiers are generated. Defining reference identifiers is the responsibility of applications or protocols that use this document. Because the security of a system that uses this document will depend on how reference identifiers are generated, great care should be taken in this process. For example, a protocol or application could specify that the application service type is obtained through a one-to-one mapping of URI schemes to service types or support only a restricted set of URI schemes. Similarly, it could insist that a domain name or IP address taken as input to the reference identifier must be obtained in a secure context such as a hyperlink embedded in a web page that was delivered over an authenticated and encrypted channel (see for instance [SECURE-CONTEXTS] with regard to the web platform).¶
Naturally, if the inputs themselves are invalid or corrupt (e.g., a user has clicked a hyperlink provided by a malicious entity in a phishing attack), then the client might end up communicating with an unexpected application service.¶
During the course of processing, a client might be exposed to identifiers that look like but are not reference identifiers. For example, DNS resolution that starts at a DNS-ID reference identifier might produce intermediate domain names that need to be further resolved. Any intermediate values are not reference identifiers and MUST NOT be treated as such. In the DNS case, not treating intermediate domain names as reference identifiers removes DNS and DNS resolution from the attack surface. However, an application might define a process for authenticating these intermediate identifiers in a way that then allows them to be used as a reference identifier; see for example [SMTP-TLS].¶
As one example of the process of generating a reference identifier, from user
input of the URI <sip:alice@example.net> a client could derive the application
service type sip
from the URI scheme and parse the domain name example.net
from the host component.¶
Using the combination of FQDN(s) or IP address(es), plus optionally an application service type, the client MUST construct its list of reference identifiers in accordance with the following rules:¶
Which identifier types a client includes in its list of reference identifiers, and their priority, is a matter of local policy. For example, a client that is built to connect only to a particular kind of service might be configured to accept as valid only certificates that include an SRV-ID for that application service type. By contrast, a more lenient client, even if built to connect only to a particular kind of service, might include SRV-IDs, DNS-IDs, and IP-IDs in its list of reference identifiers.¶
The following examples are for illustrative purposes only and are not intended to be comprehensive.¶
https://www.example.com/
would have a single reference identifier:
a DNS-ID of www.example.com
.¶
https://192.0.2.107/
would have a single
IP-ID reference identifier of 192.0.2.107
.¶
example.net
(resolved as mail.example.net
) might have three reference
identifiers: an SRV-ID of _imaps.example.net
(see [EMAIL-SRV]), and
DNS-IDs of example.net
and mail.example.net
. An email user agent that
does not support [EMAIL-SRV] would probably be explicitly configured to
connect to mail.example.net
, whereas an SRV-aware user agent would derive
example.net
from an email address of the form user@example.net
but might
also accept mail.example.net
as the DNS domain name portion of reference
identifiers for the service.¶
voice.example.edu
might have only one reference identifier:
a URI-ID of sip:voice.example.edu
(see [SIP-CERTS]).¶
im.example.org
might have three reference identifiers: an
SRV-ID of _xmpp-client.im.example.org
(see [XMPP]), a DNS-ID of
im.example.org
, and an XMPP-specific XmppAddr
of im.example.org
(see [XMPP]).¶
In all these cases, presented identifiers that do not match the reference identifier(s) would be rejected; for instance:¶
Once the client has constructed its list of reference identifiers and has received the server's presented identifiers, the client checks its reference identifiers against the presented identifiers for the purpose of finding a match. The search fails if the client exhausts its list of reference identifiers without finding a match. The search succeeds if any presented identifier matches one of the reference identifiers, at which point the client SHOULD stop the search.¶
Before applying the comparison rules provided in the following sections, the client might need to split the reference identifier into components. Each reference identifier produces either a domain name or an IP address and optionally an application service type as follows:¶
_imaps.example.net
has a DNS domain name portion
of example.net
and an application service type portion of
imaps
, which maps to the IMAP application protocol as explained in
[EMAIL-SRV].¶
sip:voice.example.edu
would be split
into a DNS domain name portion of voice.example.edu
and an application
service type of sip
(associated with an application protocol of SIP as
explained in [SIP-CERTS]).¶
If the reference identifier produces a domain name, the client MUST match the DNS name; see Section 6.3. If the reference identifier produces an IP address, the client MUST match the IP address; see Section 6.4. If an application service type is present it MUST also match the service type as well; see Section 6.5.¶
This section describes how the client must determine if the presented DNS name matches the reference DNS name. The rules differ depending on whether the domain to be checked is a traditional domain name or an internationalized domain name, as defined in Section 2. For clients that support presented identifiers containing the wildcard character "*", this section also specifies a supplemental rule for such "wildcard certificates". This section uses the description of labels and domain names in [DNS-CONCEPTS].¶
If the DNS domain name portion of a reference identifier is a "traditional
domain name" (i.e., a FQDN that conforms to "preferred name syntax" as
described in Section 3.5 of [DNS-CONCEPTS]),
then matching of the reference identifier against the presented
identifier MUST be performed by comparing the set of domain name labels using
a case-insensitive ASCII comparison, as clarified by [DNS-CASE]. For
example, WWW.Example.Com
would be lower-cased to www.example.com
for
comparison purposes. Each label MUST match in order for the names to be
considered to match, except as supplemented by the rule about checking of
wildcard labels in presented identifiers given below.¶
If the DNS domain name portion of a reference identifier is an internationalized domain name, then the client MUST convert any U-labels [IDNA-DEFS] in the domain name to A-labels before checking the domain name or comparing it with others. In accordance with [IDNA-PROTO], A-labels MUST be compared as case-insensitive ASCII. Each label MUST match in order for the domain names to be considered to match, except as supplemented by the rule about checking of wildcard labels in presented identifiers given below.¶
If the technology specification supports wildcards in presented identifiers, then the client MUST match the reference identifier against a presented identifier whose DNS domain name portion contains the wildcard character "*" in a label provided these requirements are met:¶
If the requirements are not met, the presented identifier is invalid and MUST be ignored.¶
A wildcard in a presented identifier can only match exactly one label in a reference identifier. This specification covers only wildcard characters in presented identifiers, not wildcard characters in reference identifiers or in DNS domain names more generally. Therefore the use of wildcard characters as described herein is not to be confused with DNS wildcard matching, where the "*" label always matches at least one whole label and sometimes more; see [DNS-CONCEPTS], Section 4.3.3 and [DNS-WILDCARDS].¶
For information regarding the security characteristics of wildcard certificates, see Section 7.1.¶
An IP-ID matches based on an octet-for-octet comparison of the bytes of the reference identity with the bytes contained in the iPAddress subjectAltName.¶
For an IP address that appears in a URI-ID, the "host" component of both the reference identity and the presented identifier must match. These are parsed as either an "IP-literal" (following [IPv6]) or an "IPv4address" (following [IPv4]). If the resulting octets are equal, the IP address matches.¶
This document does not specify how an SRV-ID reference identity can include an IP address.¶
The rules for matching the application service type depend on whether the identifier is an SRV-ID or a URI-ID.¶
These identifiers provide an application service type portion to be checked,
but that portion is combined only with the DNS domain name portion of the
SRV-ID or URI-ID itself. For example, if a client's list of reference
identifiers includes an SRV-ID of _xmpp-client.im.example.org
and a DNS-ID
of apps.example.net
, the client MUST check both the combination of an
application service type of xmpp-client
and a DNS domain name of
im.example.org
and, separately,
a DNS domain name of apps.example.net
. However, the
client MUST NOT check the combination of an application service type of
xmpp-client
and a DNS domain name of apps.example.net
because it does not
have an SRV-ID of _xmpp-client.apps.example.net
in its list of reference
identifiers.¶
If the identifier is an SRV-ID, then the application service name MUST
be matched in a case-insensitive manner, in accordance with [DNS-SRV].
Note that, per [SRVNAME], the _
character is part of the service name in
DNS SRV records and in SRV-IDs.¶
If the identifier is a URI-ID, then the scheme name portion MUST be
matched in a case-insensitive manner, in accordance with [URI].
Note that the :
character is a separator between the scheme name
and the rest of the URI, and thus does not need to be included in any
comparison.¶
If the client has found a presented identifier that matches a reference identifier, then the service identity check has succeeded. In this case, the client MUST use the matched reference identifier as the validated identity of the application service.¶
If the client does not find a presented identifier matching any of the reference identifiers, then the client MUST proceed as described as follows.¶
If the client is an automated application, then it SHOULD terminate the communication attempt with a bad certificate error and log the error appropriately. The application MAY provide a configuration setting to disable this behavior, but it MUST NOT disable this security control by default.¶
If the client is one that is directly controlled by a human user, then it SHOULD inform the user of the identity mismatch and automatically terminate the communication attempt with a bad certificate error in order to prevent users from inadvertently bypassing security protections in hostile situations. Such clients MAY give advanced users the option of proceeding with acceptance despite the identity mismatch. Although this behavior can be appropriate in certain specialized circumstances, it needs to be handled with extreme caution, for example by first encouraging even an advanced user to terminate the communication attempt and, if they choose to proceed anyway, by forcing the user to view the entire certification path before proceeding.¶
The application MAY also present the user with the ability to accept the presented certificate as valid for subsequent connections. Such ad-hoc "pinning" SHOULD NOT restrict future connections to just the pinned certificate. Local policy that statically enforces a given certificate for a given peer SHOULD be made available only as prior configuration, rather than a just-in-time override for a failed connection.¶
Wildcard certificates automatically vouch for any single-label host names within their domain, but not multiple levels of domains. This can be convenient for administrators but also poses the risk of vouching for rogue or buggy hosts. See for example [Defeating-SSL] (beginning at slide 91) and [HTTPSbytes] (slides 38-40).¶
As specified in Section 6.3, restricting the presented identifiers in certificates to only one wildcard character (e.g., "*.example.com" but not "*.*.example.com") and restricting the use of wildcards to only the left-most domain label can help to mitigate certain aspects of the attack described in [Defeating-SSL].¶
That same attack also relies on the initial use of a cleartext HTTP connection, which is hijacked by an active on-path attacker and subsequently upgraded to HTTPS. In order to mitigate such an attack, administrators and software developers are advised to follow the strict TLS guidelines provided in [TLS-REQS], Section 3.2.¶
Because the attack described in [HTTPSbytes] relies on an underlying cross-site scripting (XSS) attack, web browsers and applications are advised to follow best practices to prevent XSS attacks; see for example [XSS] published by the Open Web Application Security Project (OWASP).¶
Protection against a wildcard that identifies a public suffix
[Public-Suffix], such as *.co.uk
or *.com
, is beyond the scope of this
document.¶
As noted in Section 3, application protocols can disallow the use of wildcard certificates entirely as a more foolproof mitigation.¶
The URI-ID type is a subjectAltName entry of type uniformResourceIdentifier as defined in [PKIX]. For the purposes of this specification, the URI-ID MUST include both a "scheme" and a "host" component that matches the "reg-name" rule; if the entry does not include both, it is not a valid URI-ID and MUST be ignored. Any other components are ignored, because only the "scheme" and "host" components are used for certificate matching as specified under Section 6.¶
The quoted component names in the previous paragraph represent the associated [ABNF] productions from the IETF standard for Uniform Resource Identifiers [URI]. Although the reader should be aware that some applications (e.g., web browsers) might instead conform to the Uniform Resource Locator (URL) specification maintained by the WHATWG [URL], it is not expected that differences between the URI and URL specifications would manifest themselves in certificate matching.¶
This document specifies only matching between reference identifiers and presented identifiers, not the visual presentation of domain names. More specifically, matching of internationalized domain names is performed on A-labels only Section 6. The limited scope of this specification likely mitigates potential confusion caused by the use of visually similar characters in domain names (as described for example in [IDNA-DEFS], Section 4.4, [UTS-36], and [UTS-39]); in any case, such concerns are a matter for application-level protocols and user interfaces, not the matching of certificates.¶
The TLS Server Name Indication (SNI) extension only conveys domain names. Therefore, a client with an IP-ID reference identity cannot present any information about its reference identity when connecting to a server. Servers that wish to present an IP-ID therefore need to present this identity when a connection is made without SNI.¶
The textual representation of an IPv4 address might be misinterpreted as a valid FQDN in some contexts. This can result in different security treatment that might cause different components of a system to classify the value differently, which might lead to vulnerabilities. For example, one system component enforces a security rule that is conditional on the type of identifier. This component misclassifies an IP address as an FQDN. A different component correctly classifies the identifier but might incorrectly assume that rules regarding IP addresses have been enforced. Consistent classification of identifiers avoids this problem.¶
A given application service might be addressed by multiple DNS domain names for a variety of reasons, and a given deployment might service multiple domains or protocols. TLS Extensions such as TLS Server Name Indication (SNI), discussed in [TLS], Section 4.4.2.2, and Application Layer Protocol Negotiation (ALPN), discussed in [ALPN], provide a way for the application to indicate the desired identifier and protocol to the server, which it can then use to select the most appropriate certificate.¶
This specification allows multiple DNS-IDs, IP-IDs, SRV-IDs, or URI-IDs in a certificate. As a result, an application service can use the same certificate for multiple hostnames, such as when a client does not support the TLS SNI extension, or for multiple protocols, such as SMTP and HTTP, on a single hostname. Note that the set of names in a certificate is the set of names that could be affected by a compromise of any other server named in the set: the strength of any server in the set of names is determined by the weakest of those servers that offer the names.¶
Methods for mitigating this risk includes: limiting the number of names that any server can speak for, following the guidelines for use of [ALPN] described in Section 3.8 of [TLS-REQS]), and ensuring that all servers in the set have a strong minimum configuration as described in Section 3.9 of [TLS-REQS].¶
This specification describes how a client may construct multiple acceptable reference identifiers and may match any of those reference identifiers with the set of presented identifiers. [PKIX], Section 4.2.1.10 describes a mechanism to allow CA certificates to be constrained in the set of presented identifiers that they may include within server certificates. However, these constraints only apply to the explicitly enumerated name forms. For example, a CA that is only name constrained for DNS-IDs is not constrained for SRV-IDs and URI-IDs, unless those name forms are also explicitly included within the name constraints extension.¶
A client that constructs multiple reference identifiers of different types, such as both DNS-ID and SRV-IDs, as described in Section 6.1.1, SHOULD take care to ensure that CAs issuing such certificates are appropriately constrained. This MAY take the form of local policy through agreement with the issuing CA, or MAY be enforced by the client requiring that if one form of presented identifier is constrained, such as a dNSName name constraint for DNS-IDs, then all other forms of acceptable reference identities are also constrained, such as requiring a uniformResourceIndicator name constraint for URI-IDs.¶
This document has no actions for IANA.¶
This document revises and obsoletes [VERIFY] based on the decade of experience and changes since it was published. The major changes, in no particular order, include:¶
CN-ID
in [VERIFY].¶
Jeff Hodges co-authored the previous version of these recommendations, [VERIFY]. The authors gratefully acknowledge his essential contributions to this work.¶
Martin Thomson contributed the text on handling of IP-IDs.¶
We gratefully acknowledge everyone who contributed to the previous
version of these recommendations, [VERIFY].
Thanks also to Carsten Bormann for converting the previous document
to Markdown so that we could more easily use Martin Thomson's i-d-template
software.¶
In addition to discussion on the mailing list, the following people provided especially helpful feedback: Viktor Dukhovni, Jim Fenton, Olle Johansson, John Klensin, John Mattson, Alexey Melnikov, Yaron Sheffer, Ryan Sleevi, Brian Smith, and Martin Thomson.¶