RFC 9162 | Certificate Transparency Version 2.0 | December 2021 |
Laurie, et al. | Experimental | [Page] |
This document describes version 2.0 of the Certificate Transparency (CT)protocol for publicly logging the existence of Transport Layer Security (TLS)server certificates as they are issued or observed, in a manner that allowsanyone to audit certification authority (CA) activity and notice the issuance ofsuspect certificates as well as to audit the certificate logs themselves. Theintent is that eventually clients would refuse to honor certificates that do notappear in a log, effectively forcing CAs to add all issued certificates to thelogs.¶
This document obsoletes RFC 6962. It also specifies a new TLS extension that isused to send various CT log artifacts.¶
Logs are network services that implement the protocol operations for submissionsand queries that are defined in this document.¶
This document is not an Internet Standards Track specification; it is published for examination, experimental implementation, and evaluation.¶
This document defines an Experimental Protocol for the Internet community. 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 candidates for any level of Internet Standard; see Section 2 of RFC 7841.¶
Information about the current status of this document, any errata, and how to provide feedback on it may be obtained athttps://www.rfc-editor.org/info/rfc9162.¶
Copyright (c) 2021 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.¶
Certificate Transparency aims to mitigate the problem of misissued certificatesby providing append-only logs of issued certificates. The logs do not themselvesprevent misissuance, but they ensure that interested parties (particularly thosenamed in certificates) can detect such misissuance. Note that this is a generalmechanism that could be used for transparently logging any form of binary data,subject to some kind of inclusion criteria. In this document, we only describeits use for public TLS server certificates (i.e., where the inclusion criteriais a valid certificate issued by a public certification authority (CA)).A typical definition of "public" can be found in[CABBR].¶
Each log contains certificate chains, which can be submitted by anyone. It isexpected that public CAs will contribute all their newly issued certificates toone or more logs; however, certificate holders can also contribute their owncertificate chains, as can third parties. In order to avoid logs being rendereduseless by the submission of large numbers of spurious certificates, it isrequired that each chain ends with a trust anchor that is accepted by the log.A log may also limit the length of the chain it is willing to accept;such chains must also end with an acceptable trust anchor.When a chain is accepted by a log, a signed timestamp is returned, which canlater be used to provide evidence to TLS clients that the chain has beensubmitted. TLS clients can thus require that all certificates they accept asvalid are accompanied by signed timestamps.¶
Those who are concerned about misissuance can monitor the logs, asking themregularly for all new entries, and can thus check whether domains for which theyare responsible have had certificates issued that they did not expect. Whatthey do with this information, particularly when they find that a misissuancehas happened, is beyond the scope of this document. However, broadly speaking,they can invoke existing business mechanisms for dealing with misissuedcertificates, such as working with the CA to get the certificate revoked orwith maintainers of trust anchor lists to get the CA removed. Of course, anyonewho wants can monitor the logs and, if they believe a certificate is incorrectlyissued, take action as they see fit.¶
Similarly, those who have seen signed timestamps from a particular log can laterdemand a proof of inclusion from that log. If the log is unable to provide this(or, indeed, if the corresponding certificate is absent from monitors' copies ofthat log), that is evidence of the incorrect operation of the log. The checkingoperation is asynchronous to allow clients to proceed without delay, despitepossible issues, such as network connectivity and the vagaries of firewalls.¶
The append-only property of each log is achieved using Merkle Trees, which canbe used to efficiently prove that any particular instance of the log is asuperset of any particular previous instance and to efficiently detect variousmisbehaviors of the log (e.g., issuing a signed timestamp for a certificate thatis not subsequently logged).¶
The log auditing mechanisms described in this document canbe circumvented by a misbehaving log that shows different, inconsistentviews of itself to different clients. Therefore, it is necessary to treat each log as a trusted third party. While mechanisms are being developed to address theseshortcomings and thereby avoid the need to blindly trust logs,such mechanisms are outside the scope of this document.¶
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.¶
Data structures are defined and encoded according to the conventions laid out inSection 3 of [RFC8446].¶
This document uses object identifiers (OIDs) to identify Log IDs (seeSection 4.4), the precertificate Cryptographic Message Syntax (CMS)eContentType
(seeSection 3.2), X.509v3 extensions in certificates (seeSection 7.1.2), and Online Certificate Status Protocol (OCSP) responses (seeSection 7.1.1). The OIDs are defined in an arc that was selected due to its short encoding.¶
This document revises and obsoletes the CT 1.0 protocol[RFC6962], drawing on insights gained from CT 1.0 deployments and on feedback from the community. The major changes are:¶
TransItem
structure: This new data structure is used to encapsulate most types of CT data. ATransItemList
, consisting of one or moreTransItem
structures, can be used anywhere thatSignedCertificateTimestampList
was used in[RFC6962].¶MerkleTreeLeaf
structure has been replaced by theTransItem
structure, which eases extensibility and simplifies the leaf structure by removing one layer of abstraction.¶TransItem
structures are returned rather than the constituent parts of each structure.¶get-all-by-hash
: This is a new client API for obtaining an inclusion proof and the corresponding consistency proof at the same time.¶submit-entry
: This is a new client API, replacingadd-chain
andadd-pre-chain
.¶signed_certificate_timestamp
TLS extension has been replaced by thetransparency_info
TLS extension.¶A full description of the Merkle Tree is beyond the scope of this document. Briefly, it is a binary tree where each non-leaf node is a hash of its children. For CT, the number of children is at most two. Additional information can be found in the Introduction and Reference sections of[RFC8391].¶
The log uses a binary Merkle Tree for efficient auditing. The hash algorithm used is one of the log's parameters (seeSection 4.1). This document establishes a registry of acceptable hash algorithms (seeSection 10.2.1). Throughout this document, the hash algorithm in use is referred to as HASH and the size of its output in bytes is referred to as HASH_SIZE. The input to the Merkle Tree Hash is a list of data entries; these entries will be hashed to form the leaves of the Merkle Tree. The output is a single HASH_SIZE Merkle Tree Hash. Given an ordered list of n inputs, D_n = {d[0], d[1], ..., d[n-1]}, the Merkle Tree Hash (MTH) is thus defined as follows:¶
The hash of an empty list is the hash of an empty string:¶
MTH({}) = HASH().¶
The hash of a list with one entry (also known as a leaf hash) is:¶
MTH({d[0]}) = HASH(0x00 || d[0]).¶
For n > 1, let k be the largest power of two smaller than n (i.e., k < n <= 2k). The Merkle Tree Hash of an n-element list D_n is then defined recursively as:¶
MTH(D_n) = HASH(0x01 || MTH(D[0:k]) || MTH(D[k:n])),¶
where:¶
Note that the hash calculations for leaves and nodes differ; this domainseparation is required to give second preimage resistance.¶
Note that we do not require the length of the input list to be a power of two.The resulting Merkle Tree may thus not be balanced; however, its shape isuniquely determined by the number of leaves. (Note: This Merkle Tree isessentially the same as the history tree proposed by[CrosbyWallach], except ourdefinition handles non-full trees differently.)¶
When a client has a complete list ofentries
from0
up totree_size - 1
and wishes to verify this list against a tree headroot_hash
returned by the log for the sametree_size
, the following algorithm may beused:¶
stack
to an empty stack.¶For eachi
from0
up totree_size - 1
:¶
HASH(0x00 || entries[i])
tostack
.¶merge_count
to the lowest value (0
included) such thatLSB(i >> merge_count)
is not set, whereLSB
means the least significant bit. In other words, setmerge_count
to the number of consecutive1
s found starting at the least significant bit ofi
.¶Repeatmerge_count
times:¶
stack
, repeat the same mergeprocedure (the sub-items of Step 2(c) above) until only a single elementremains.¶stack
is the Merkle Tree Hash for the giventree_size
and should be compared by equality against the suppliedroot_hash
.¶A Merkle inclusion proof for a leaf in a Merkle Tree is the shortest listof additional nodes in the Merkle Tree required to compute the Merkle Tree Hashfor that tree. Each node in the tree is either a leaf node or is computed fromthe two nodes immediately below it (i.e., towards the leaves). At each step upthe tree (towards the root), a node from the inclusion proof is combined withthe node computed so far. In other words, the inclusion proof consists of thelist of missing nodes required to compute the nodes leading from a leaf to theroot of the tree. If the root computed from the inclusion proof matches the trueroot, then the inclusion proof proves that the leaf exists in the tree.¶
Given an ordered list of n inputs to the tree, D_n = {d[0], d[1], ...,d[n-1]}, the Merkle inclusion proof PATH(m, D_n) for the (m+1)th input d[m],0 <= m < n, is defined as follows:¶
The proof for the single leaf in a tree with a one-element input list D[1] ={d[0]} is empty:¶
PATH(0, {d[0]}) = {}¶
For n > 1, let k be the largest power of two smaller than n. The proof for the(m+1)th element d[m] in a list of n > m elements is then defined recursively as:¶
PATH(m, D_n) = PATH(m, D[0:k]) : MTH(D[k:n]) for m < k; andPATH(m, D_n) = PATH(m - k, D[k:n]) : MTH(D[0:k]) for m >= k,¶
The : operator and D[k1:k2] are defined the same as inSection 2.1.1.¶
When a client has received an inclusion proof (e.g., in aTransItem
of typeinclusion_proof_v2
) and wishes to verify inclusion of an inputhash
for agiventree_size
androot_hash
, the following algorithm may be used to provethehash
was included in theroot_hash
:¶
leaf_index
from theinclusion_proof_v2
structure againsttree_size
. Ifleaf_index
is greater than or equal totree_size
, then fail the proof verification.¶fn
toleaf_index
andsn
totree_size - 1
.¶r
tohash
.¶For each valuep
in theinclusion_path
array:¶
sn
is 0, then stop the iteration and fail the proof verification.¶IfLSB(fn)
is set, or iffn
is equal tosn
, then:¶
r
toHASH(0x01 || p || r)
.¶LSB(fn)
is not set, then right-shift bothfn
andsn
equally until eitherLSB(fn)
is set orfn
is0
.¶Otherwise:¶
r
toHASH(0x01 || r || p)
.¶fn
andsn
one time.¶sn
to 0. Comparer
against theroot_hash
. Ifsn
is equal to 0 andr
and theroot_hash
are equal, then the log has proven the inclusion ofhash
. Otherwise, fail the proof verification.¶Merkle consistency proofs prove the append-only property of the tree. A Merkleconsistency proof for a Merkle Tree Hash MTH(D_n) and a previously advertisedhash MTH(D[0:m]) of the first m leaves, m <= n, is the list of nodes in theMerkle Tree required to verify that the first m inputs D[0:m] are equal in bothtrees. Thus, a consistency proof must contain a set of intermediate nodes (i.e.,commitments to inputs) sufficient to verify MTH(D_n), such that (a subset of)the same nodes can be used to verify MTH(D[0:m]). We define an algorithm thatoutputs the (unique) minimal consistency proof.¶
Given an ordered list of n inputs to the tree, D_n = {d[0], d[1], ...,d[n-1]}, the Merkle consistency proof PROOF(m, D_n) for a previous Merkle TreeHash MTH(D[0:m]), 0 < m < n, is defined as:¶
PROOF(m, D_n) = SUBPROOF(m, D_n, true)¶
In SUBPROOF, the boolean value represents whether the subtree created from D[0:m] is a complete subtree of the Merkle Tree created from D_n and, consequently, whether the subtree Merkle Tree Hash MTH(D[0:m]) is known. The initial call to SUBPROOF sets this to be true, and SUBPROOF is then defined as follows:¶
The subproof for m = n is empty if m is the value for which PROOF was originally requested (meaning that the subtree created from D[0:m] is a complete subtree of the Merkle Tree created from the original D_n for which PROOF was requested and the subtree Merkle Tree Hash MTH(D[0:m]) is known):¶
SUBPROOF(m, D_m, true) = {}¶
Otherwise, the subproof for m = n is the Merkle Tree Hash committing inputs D[0:m]:¶
SUBPROOF(m, D_m, false) = {MTH(D_m)}¶
For m < n, let k be the largest power of two smaller than n. The subproof is then defined recursively, using the appropriate step below:¶
If m <= k, the right subtree entries D[k:n] only exist in the current tree. We prove that the left subtree entries D[0:k] are consistent and add a commitment to D[k:n]:¶
SUBPROOF(m, D_n, b) = SUBPROOF(m, D[0:k], b) : MTH(D[k:n])¶
If m > k, the left subtree entries D[0:k] are identical in both trees. We prove that the right subtree entries D[k:n] are consistent and add a commitment to D[0:k]:¶
SUBPROOF(m, D_n, b) = SUBPROOF(m - k, D[k:n], false) : MTH(D[0:k])¶
The number of nodes in the resulting proof is bounded above by ceil(log2(n)) +1.¶
The : operator and D[k1:k2] are defined the same as inSection 2.1.1.¶
When a client has a tree headfirst_hash
for tree sizefirst
, has a tree headsecond_hash
for tree sizesecond
where0 < first < second
, and has received a consistency proof between the two (e.g., in aTransItem
of typeconsistency_proof_v2
), the following algorithm may be used to verify the consistency proof:¶
consistency_path
is an empty array, stop and fail the proof verification.¶first
is an exact power of 2, then prependfirst_hash
to theconsistency_path
array.¶fn
tofirst - 1
andsn
tosecond - 1
.¶LSB(fn)
is set, then right-shift bothfn
andsn
equally untilLSB(fn)
is not set.¶fr
andsr
to the first value in theconsistency_path
array.¶For each subsequent valuec
in theconsistency_path
array:¶
sn
is 0, then stop the iteration and fail the proof verification.¶IfLSB(fn)
is set, or iffn
is equal tosn
, then:¶
fr
toHASH(0x01 || c || fr)
.¶sr
toHASH(0x01 || c || sr)
.¶LSB(fn)
is not set, then right-shift bothfn
andsn
equally until eitherLSB(fn)
is set orfn
is0
.¶Otherwise:¶
sr
toHASH(0x01 || sr || c)
.¶fn
andsn
one time.¶consistency_path
array as described above, verify that thefr
calculated is equal to thefirst_hash
supplied, that thesr
calculated is equal to thesecond_hash
supplied, and thatsn
is 0.¶The following is a binary Merkle Tree with 7 leaves:¶
hash / \ / \ / \ / \ / \ k l / \ / \ / \ / \ / \ / \ g h i j / \ / \ / \ | a b c d e f d6 | | | | | |d0 d1 d2 d3 d4 d5¶
The inclusion proof ford0
is [b
,h
,l
].¶
The inclusion proof ford3
is [c
,g
,l
].¶
The inclusion proof ford4
is [f
,j
,k
].¶
The inclusion proof ford6
is [i
,k
].¶
The same tree, built incrementally in four steps:¶
hash0 hash1=k / \ / \ / \ / \ / \ / \ g c g h / \ | / \ / \ a b d2 a b c d | | | | | |d0 d1 d0 d1 d2 d3 hash2 hash / \ / \ / \ / \ / \ / \ / \ / \ / \ / \ k i k l / \ / \ / \ / \ / \ e f / \ / \ / \ | | / \ / \ g h d4 d5 g h i j/ \ / \ / \ / \ / \ |a b c d a b c d e f d6| | | | | | | | | |d0 d1 d2 d3 d0 d1 d2 d3 d4 d5¶
The consistency proof betweenhash0
andhash
is PROOF(3, D[7]) = [c
,d
,g
,l
].Non-leaf nodesc
,g
are used to verifyhash0
, and non-leaf nodesd
,l
are additionally used to showhash
isconsistent withhash0
.¶
The consistency proof betweenhash1
andhash
is PROOF(4, D[7]) = [l
].hash
canbe verified usinghash1
=k
andl
.¶
The consistency proof betweenhash2
andhash
is PROOF(6, D[7]) = [i
,j
,k
].Non-leaf nodesk
,i
are used to verifyhash2
, and non-leaf nodej
is additionally used to showhash
isconsistent withhash2
.¶
When signing data structures, a logMUST use one ofthe signature algorithms from the IANA "Signature Algorithms" registry,described inSection 10.2.2.¶
Submitters submit certificates or preannouncements of certificates prior toissuance (precertificates) to logs for public auditing, as described below. Inorder to enable attribution of each logged certificate or precertificate to itsissuer, each submissionMUST be accompanied by all additional certificatesrequired to verify the chain up to an accepted trust anchor (Section 5.7).The trust anchor (a root or intermediate CA certificate)MAY be omitted from thesubmission.¶
If a log accepts a submission, it will return a Signed Certificate Timestamp(SCT) (seeSection 4.8). The submitterSHOULD validate the returned SCT, as describedinSection 8.1, if they understand its format and they intend to use itdirectly in a TLS handshake or to construct a certificate. If the submitter doesnot need the SCT (for example, the certificate is being submitted simply to makeit available in the log), itMAY validate the SCT.¶
Any entity can submit a certificate (Section 5.1) to a log. Since it isanticipated that TLS clients will reject certificates that are not logged, it isexpected that certificate issuers and subjects will be strongly motivated tosubmit them.¶
CAs may preannounce a certificate prior to issuance by submitting aprecertificate (Section 5.1) that the log can use to create an entry thatwill be valid against the issued certificate. The CAMAY incorporate thereturned SCT in the issued certificate. One example of where the returned SCT isnot incorporated in the issued certificate is when a CA sends the precertificateto multiple logs but only incorporates the SCTs that are returned first.¶
A precertificate is a CMS[RFC5652]signed-data
object that conforms to thefollowing profile:¶
SignedData.version
MUST be v3(3).¶SignedData.digestAlgorithms
MUST be the same as theSignerInfo.digestAlgorithm
OID value (see below).¶SignedData.encapContentInfo
:¶
eContentType
MUST be the OID 1.3.101.78.¶eContent
MUST contain a TBSCertificate[RFC5280] that will be identical to the TBSCertificate in the issued certificate, except that the Transparency Information (Section 7.1) extensionMUST be omitted.¶SignedData.certificates
MUST be omitted.¶SignedData.crls
MUST be omitted.¶SignedData.signerInfos
MUST contain oneSignerInfo
:¶
version
MUST be v3(3).¶sid
MUST use thesubjectKeyIdentifier
option.¶digestAlgorithm
MUST be one of the hash algorithm OIDs listed in the IANA "Hash Algorithms" registry, described inSection 10.2.1.¶signedAttrs
MUST be present andMUST contain two attributes:¶
signatureAlgorithm
MUST be the same OID asTBSCertificate.signature
.¶signature
MUST be from the same (root or intermediate) CA that intends to issue the corresponding certificate (seeSection 3.2.1).¶unsignedAttrs
MUST be omitted.¶SignerInfo.signedAttrs
is included in the message digest calculation process(seeSection 5.4 of [RFC5652]), which ensures that theSignerInfo.signature
value will not be a valid X.509v3 signature that could be used in conjunctionwith the TBSCertificate (fromSignedData.encapContentInfo.eContent
) toconstruct a valid certificate.¶
Under normal circumstances, there will be a short delay between precertificatesubmission and issuance of the corresponding certificate. Longer delays are tobe expected occasionally (e.g., due to log server downtime); in some cases,the CA might not actually issue the corresponding certificate. Nevertheless, aprecertificate'ssignature
indicates the CA's binding intent to issue thecorresponding certificate, which means that:¶
A log is a single, append-only Merkle Tree of submitted certificate andprecertificate entries.¶
When it receives and accepts a valid submission, the logMUST return an SCT thatcorresponds to the submitted certificate or precertificate. If the log haspreviously seen this valid submission, itSHOULD return the same SCT as itreturned before, as discussed inSection 11.3.If different SCTs are produced for the samesubmission, multiple log entries will have to be created, one for each SCT (asthe timestamp is a part of the leaf structure). Note that if a certificate waspreviously logged as a precertificate, then the precertificate's SCT of typeprecert_sct_v2
would not be appropriate; instead, a fresh SCT of typex509_sct_v2
should be generated.¶
An SCT is the log's promise to append to its Merkle Tree an entry for the accepted submission. Upon producing an SCT, the logMUST fulfill this promise by performing the following actions within a fixed amount of time known as the Maximum Merge Delay (MMD), which is one of the log's parameters (seeSection 4.1):¶
The log may append multiple entries before signing the root of the tree.¶
Log operatorsSHOULD NOT impose any conditions on retrieving or sharing datafrom the log.¶
A log is defined by a collection of immutable parameters, which are used byclients to communicate with the log and to verify log artifacts. Except for theFinal STH, each of these parametersMUST be establishedbefore the log operator begins to operate the log.¶
Maximum Merge Delay
(seeSection 4.10).¶TransItem
of typesigned_tree_head_v2
. If a log is still accepting entries, this value should not be provided.¶[JSON.Metadata] is an example of a metadata format that includes the above elements.¶
A log determines whether to accept or reject a submission by evaluating itagainst the minimum acceptance criteria (seeSection 4.2.1) and againstthe log's discretionary acceptance criteria (seeSection 4.2.2).¶
If the acceptance criteria are met, the logSHOULD accept the submission. (A logmay decide, for example, to temporarily reject acceptable submissions to protectitself against denial-of-service attacks.)¶
The logSHALL allow retrieval of its list of accepted trust anchors (seeSection 5.7), each of which is a root or intermediate CA certificate. Thislist might usefully be the union of root certificates trusted by major browservendors.¶
To ensure that logged certificates and precertificates are attributable to anaccepted trust anchor, to set clear expectations for what monitors wouldfind in the log, and to avoid being overloaded by invalid submissions, the logMUST reject a submission if any of the following conditions are not met:¶
submission
,type
, andchain
inputsMUST be set as described inSection 5.1. The logMUST NOT accommodate misordered CA certificates or use any other source of intermediate CA certificates to attempt certification path construction.¶Each of the zero or more intermediate CA certificates in the chainMUST have one or both of the following features:¶
If the minimum acceptance criteria are met but the submission is not fully valid according to[RFC5280] verification rules (e.g., the certificate or precertificate has expired, is not yet valid, has been revoked, exhibits ASN.1 DER encoding errors but the log can still parse it, etc.), then the acceptability of the submission is left to the log's discretion. It is useful for logs to accept such submissions in order to accommodate quirks of CA certificate-issuing software and to facilitate monitoring of CA compliance with applicable policies and technical standards. However, it is impractical for this document to enumerate, and for logs to consider, all of the ways that a submission might fail to comply with[RFC5280].¶
LogsSHOULD limit the length of chain they will accept. The maximum chain length is one of the log's parameters (seeSection 4.1).¶
If a submission is accepted and an SCT is issued, the accepting logMUST store theentire chain used for verification. This chainMUST include the certificate orprecertificate itself, the zero or more intermediate CA certificates provided bythe submitter, and the trust anchor used to verify the chain (even if it wasomitted from the submission). The logMUST provide this chain for auditing uponrequest (seeSection 5.6) so that the CA cannot avoid blame bylogging a partial or empty chain.Each log entry is aTransItem
structure of typex509_entry_v2
orprecert_entry_v2
. However, a log may store its entries in any format. If alog does not store thisTransItem
in full, it must store thetimestamp
andsct_extensions
of the correspondingTimestampedCertificateEntryDataV2
structure. TheTransItem
can be reconstructed from these fields and the entirechain that the log used to verify the submission.¶
Each log is identified by an OID, which is one of the log's parameters (seeSection 4.1) and whichMUST NOT be used to identify any other log. Alog's operatorMUST either allocate the OID themselves or request an OID fromthe Log ID registry (seeSection 10.2.5).One way to get an OID arc, from which OIDs can be allocated, is to requesta Private Enterprise Number from IANA by completing theregistration form.The only advantage of the registry is that the DER encoding can be small.(Recall that OID allocations do not require a central registration, althoughlogs will most likely want to make themselves known to potential clientsthrough out-of-band means.)Various data structures includethe DER encoding of this OID, excluding the ASN.1 tag and length bytes, in anopaque vector:¶
opaque LogID<2..127>;¶
Note that the ASN.1 length and the opaque vector length are identical in size (1byte) and value, so the full DER encoding (including the tag and length)of the OID can be reproduced simply byprepending an OBJECT IDENTIFIER tag (0x06) to the opaque vector length andcontents.¶
The OID used to identify a log is limited such that the DER encoding of itsvalue, excluding the tag and length,MUST be no longer than 127 octets.¶
Various data structures are encapsulated in theTransItem
structure to ensurethat the type and version of each one is identified in a common fashion:¶
enum { x509_entry_v2(0x0100), precert_entry_v2(0x0101), x509_sct_v2(0x0102), precert_sct_v2(0x0103), signed_tree_head_v2(0x0104), consistency_proof_v2(0x0105), inclusion_proof_v2(0x0106), /* Reserved Code Points */ reserved_rfc6962(0x0000..0x00FF), reserved_experimentaluse(0xE000..0xEFFF), reserved_privateuse(0xF000..0xFFFF), (0xFFFF) } VersionedTransType; struct { VersionedTransType versioned_type; select (versioned_type) { case x509_entry_v2: TimestampedCertificateEntryDataV2; case precert_entry_v2: TimestampedCertificateEntryDataV2; case x509_sct_v2: SignedCertificateTimestampDataV2; case precert_sct_v2: SignedCertificateTimestampDataV2; case signed_tree_head_v2: SignedTreeHeadDataV2; case consistency_proof_v2: ConsistencyProofDataV2; case inclusion_proof_v2: InclusionProofDataV2; } data; } TransItem;¶
versioned_type
is a value from the IANA registry inSection 10.2.3that identifies the type of the encapsulated data structure and the earliestversion of this protocol to which it conforms. This document is v2.¶
data
is the encapsulated data structure. The various structures named with theDataV2
suffix are defined in later sections of this document.¶
Note thatVersionedTransType
combines the v1 type enumerationsVersion
,LogEntryType
,SignatureType
, andMerkleLeafType
[RFC6962]. Note also thatv1 did not defineTransItem
, but this document provides guidelines (seeAppendix A) on how v2 implementations can coexist with v1implementations.¶
Future versions of this protocol may reuseVersionedTransType
values definedin this document as long as the corresponding data structures are not modifiedand may add newVersionedTransType
values for new or modified data structures.¶
enum { reserved(65535) } ExtensionType; struct { ExtensionType extension_type; opaque extension_data<0..2^16-1>; } Extension;¶
TheExtension
structure provides a generic extensibility for log artifacts,including SCTs (Section 4.8) and STHs(Section 4.10). The interpretation of theextension_data
field is determined solelyby the value of theextension_type
field.¶
This document does not define any extensions, but it does establish a registryfor futureExtensionType
values (seeSection 10.2.4).Each document that registers a newExtensionType
must specify the context inwhich it may be used (e.g., SCT, STH, or both) and describe how to interpret thecorrespondingextension_data
.¶
The leaves of a log's Merkle Tree correspond to the log's entries (seeSection 4.3). Each leaf is the leaf hash (Section 2.1) of aTransItem
structure of typex509_entry_v2
orprecert_entry_v2
, which encapsulates aTimestampedCertificateEntryDataV2
structure. Note that leaf hashes arecalculated asHASH(0x00 || TransItem)
, where the hash algorithm is one of thelog's parameters.¶
opaque TBSCertificate<1..2^24-1>; struct { uint64 timestamp; opaque issuer_key_hash<32..2^8-1>; TBSCertificate tbs_certificate; Extension sct_extensions<0..2^16-1>; } TimestampedCertificateEntryDataV2;¶
timestamp
is the date and time at which the certificate or precertificate was accepted by the log, in the form of a 64-bit unsigned number of milliseconds elapsed since the Unix Epoch (1 January 1970 00:00:00 UTC -- see[UNIXTIME]), ignoring leap seconds, in network byte order. Note that the leaves of a log's Merkle Tree are not required to be in strict chronological order.¶
issuer_key_hash
is the HASH of the public key of the CA that issued thecertificate or precertificate, calculated over the DER encoding of the keyrepresented as SubjectPublicKeyInfo[RFC5280]. This is needed to bind the CA tothe certificate or precertificate, making it impossible for the correspondingSCT to be valid for any other certificate or precertificate whose TBSCertificatematchestbs_certificate
. The length of theissuer_key_hash
MUST matchHASH_SIZE.¶
tbs_certificate
is the DER-encoded TBSCertificate from the submission. (Note that a precertificate's TBSCertificate can be reconstructed from the corresponding certificate, as described inSection 8.1.2).¶
sct_extensions
is byte-for-byte identical to the SCT extensions of the corresponding SCT.¶
The type of theTransItem
corresponds to the value of thetype
parametersupplied in theSection 5.1 call.¶
An SCT is aTransItem
structure of typex509_sct_v2
orprecert_sct_v2
,which encapsulates aSignedCertificateTimestampDataV2
structure:¶
struct { LogID log_id; uint64 timestamp; Extension sct_extensions<0..2^16-1>; opaque signature<1..2^16-1>; } SignedCertificateTimestampDataV2;¶
log_id
is this log's unique ID, encoded in an opaque vector, as described inSection 4.4.¶
timestamp
is equal to the timestamp from the correspondingTimestampedCertificateEntryDataV2
structure.¶
sct_extensions
is a vector of 0 or more SCT extensions. This vectorMUST NOT include more than one extension with the sameextension_type
. The extensions in the vectorMUST be ordered by the value of theextension_type
field, smallest value first. All SCT extensions are similar to noncritical X.509v3 extensions (i.e., themustUnderstand
field is not set), and a recipientSHOULD ignore any extension it does not understand. Furthermore, an implementationMAY choose to ignore any extension(s) that it does understand.¶
signature
is computed over aTransItem
structure of typex509_entry_v2
orprecert_entry_v2
(seeSection 4.7) using the signature algorithm declared in the log's parameters (seeSection 4.1).¶
The log stores information about its Merkle Tree in aTreeHeadDataV2
:¶
opaque NodeHash<32..2^8-1>; struct { uint64 timestamp; uint64 tree_size; NodeHash root_hash; Extension sth_extensions<0..2^16-1>; } TreeHeadDataV2;¶
The length of NodeHashMUST match HASH_SIZE of the log.¶
timestamp
is the current date and time, using the format defined inSection 4.7.¶
tree_size
is the number of entries currently in the log's Merkle Tree.¶
root_hash
is the root of the Merkle Tree.¶
sth_extensions
is a vector of 0 or more STH extensions. This vectorMUST NOTinclude more than one extension with the sameextension_type
. Theextensions in the vectorMUST be ordered by the value of theextension_type
field, smallest value first. If an implementation sees anextension that it does not understand, itSHOULD ignore that extension.Furthermore, an implementationMAY choose to ignore any extension(s) that itdoes understand.¶
Periodically, each logSHOULD sign its current tree head information (seeSection 4.9) to produce an STH. When a client requests a log's latest STH (seeSection 5.2), the logMUST return an STH that is no older than the log's MMD. However, since STHs could be used to mark individual clients (by producing a new STH for each query), a logMUST NOT produce STHs more frequently than its parameters declare (seeSection 4.1). In general, there is no need to produce a new STH unless there are new entries in the log; however, in the event that a log does not accept any submissions during an MMD period, the logMUST sign the same Merkle Tree Hash with a fresh timestamp.¶
An STH is aTransItem
structure of typesigned_tree_head_v2
, which encapsulates aSignedTreeHeadDataV2
structure:¶
struct { LogID log_id; TreeHeadDataV2 tree_head; opaque signature<1..2^16-1>; } SignedTreeHeadDataV2;¶
log_id
is this log's unique ID encoded in an opaque vector, as described inSection 4.4.¶
Thetimestamp
intree_head
MUST be at least as recent as the most recent SCT timestamp in the tree. Each subsequent timestampMUST be more recent than the timestamp of the previous update.¶
tree_head
contains the latest tree head information (seeSection 4.9).¶
signature
is computed over thetree_head
field using the signature algorithmdeclared in the log's parameters (seeSection 4.1).¶
To prepare a Merkle consistency proof for distribution to clients, the logproduces aTransItem
structure of typeconsistency_proof_v2
, whichencapsulates aConsistencyProofDataV2
structure:¶
struct { LogID log_id; uint64 tree_size_1; uint64 tree_size_2; NodeHash consistency_path<0..2^16-1>; } ConsistencyProofDataV2;¶
log_id
is this log's unique ID encoded in an opaque vector, as described inSection 4.4.¶
tree_size_1
is the size of the older tree.¶
tree_size_2
is the size of the newer tree.¶
consistency_path
is a vector of Merkle Tree nodes proving the consistency of two STHs, as described inSection 2.1.4.¶
To prepare a Merkle inclusion proof for distribution to clients, the logproduces aTransItem
structure of typeinclusion_proof_v2
, whichencapsulates anInclusionProofDataV2
structure:¶
struct { LogID log_id; uint64 tree_size; uint64 leaf_index; NodeHash inclusion_path<0..2^16-1>; } InclusionProofDataV2;¶
log_id
is this log's unique ID encoded in an opaque vector, as described inSection 4.4.¶
tree_size
is the size of the tree on which this inclusion proof is based.¶
leaf_index
is the 0-based index of the log entry corresponding to thisinclusion proof.¶
inclusion_path
is a vector of Merkle Tree nodes proving the inclusion of thechosen certificate or precertificate, as described inSection 2.1.3.¶
Log operators may decide to shut down a log for various reasons, such asdeprecation of the signature algorithm. If there are entries in the log forcertificates that have not yet expired, simply making TLS clients stoprecognizing that log will have the effect of invalidating SCTs from that log.In order to avoid that, the following actionsSHOULD be taken:¶
Messages are sent as HTTPS GET or POST requests. Parameters for POSTs and allresponses are encoded as JavaScript Object Notation (JSON) objects[RFC8259].Parameters for GETs are encoded as order-independent key/value URL parameters,using the "application/x-www-form-urlencoded" format described in the "HTML 4.01Specification"[HTML401]. Binary data is base64 encoded according toSection 4 of [RFC4648], as specifiedin the individual messages.¶
Clients are configured with a log's base URL, which is one of the log'sparameters. Clients construct URLs for requests by appending suffixes to thisbase URL. This structure places some degree of restriction on how log operatorscan deploy these services, as noted in[RFC8820]. However, operationalexperience with version 1 of this protocol has not indicated that theserestrictions are a problem in practice.¶
Note that JSON objects and URL parameters may contain fields not specified hereto allow for experimentation. Any fields that are not understoodSHOULDbe ignored.¶
In practice, log servers may include multiple front-end machines. Since it isimpractical to keep these machines in perfect sync, errors that are caused by skew between the machines may occur. Where such errors are possible, thefront end will return additional information (as specified below), making itpossible for clients to make progress, if progress is possible. Front endsMUSTonly serve data that is free of gaps (that is, for example, no front end willrespond with an STH unless it is also able to prove consistency from all logentries logged within that STH).¶
For example, when a consistency proof between two STHs is requested, thefront end reached may not yet be aware of one or both STHs. In the case where itis unaware of both, it will return the latest STH it is aware of. Where it isaware of the first but not the second, it will return the latest STH it is awareof and a consistency proof from the first STH to the returned STH. The casewhere it knows the second but not the first should not arise (see the "no gaps"requirement above).¶
If the log is unable to process a client's request, itMUST return an HTTPresponse code of 4xx/5xx (see[RFC7231]), and, in place of the responsesoutlined in the subsections below, the bodySHOULD be a JSON problem detailsobject (seeSection 3 of [RFC7807]) containing:¶
type
field within the URN namespace of: "urn:ietf:params:trans:error:".¶For example, in response to a request of<Base URL>/ct/v2/get-entries?start=100&end=99
, the log would return a400 Bad Request
response code with a body similar to the following:¶
{ "type": "urn:ietf:params:trans:error:endBeforeStart", "detail": "'start' cannot be greater than 'end'" }¶
Most error types are specific to the type of request and are defined in therespective subsections below. The one exception is the "malformed" error type,which indicates that the log server could not parse the client's request becauseit did not comply with this document:¶
type | detail |
---|---|
malformed | The request could not be parsed. |
ClientsSHOULD treat500 Internal Server Error
and503 Service Unavailable
responses as transient failures andMAY retry the same request without modification at a later date. Note that in the case of a 503 response, the logMAY include aRetry-After
header field per[RFC7231] in order to request a minimum time for the client to wait before retrying the request. In the absence of this header field, this document does not specify a minimum.¶
ClientsSHOULD treat any 4xx error as a problem with the request and not attempt to resubmit without some modification to the request. The full status codeMAY provide additional details.¶
This document deliberately does not provide more specific guidance on the use of HTTP status codes.¶
POST <Base URL>/ct/v2/submit-entry¶
VersionedTransType
integer value that indicates the type of thesubmission
: 1 forx509_entry_v2
or 2 forprecert_entry_v2
.¶submission
, the second certifies the first, etc. The last element ofchain
(or, ifchain
is an empty array, thesubmission
) is certified by an accepted trust anchor.¶A base64-encodedTransItem
of typex509_sct_v2
orprecert_sct_v2
, signed by this log, that corresponds to thesubmission
.¶
If the submitted entry is immediately appended to (or already exists in) this log's tree, then the logSHOULD also output:¶
Error codes:¶
type | detail |
---|---|
badSubmission | submission is neither a valid certificate nor a valid precertificate. |
badType | type is neither 1 nor 2. |
badChain | The first element ofchain is not the certifier of thesubmission , or the second element does not certify the first, etc. |
badCertificate | One or more certificates inchain are not valid (e.g., not properly encoded). |
unknownAnchor | The last element ofchain (or, ifchain is an empty array, thesubmission ) is not, nor is it certified by, an accepted trust anchor. |
shutdown | The log is no longer accepting submissions. |
If the version ofsct
is not v2, then a v2 client may be unable to verify thesignature. ItMUST NOT construe this as an error. This is to avoid forcing anupgrade of compliant v2 clients that do not use the returned SCTs.¶
If a log detects bad encoding in a chain that otherwise verifies correctly, thenthe logMUST either log the certificate or return the "badCertificate" error.If the certificate is logged, an SCTMUST be issued. Logging the certificate isuseful, because monitors (Section 8.2) can then detect these encoding errors,which may be accepted by some TLS clients.¶
Ifsubmission
is an accepted trust anchor whose certifier is neither anaccepted trust anchor nor the first element ofchain
, then the logMUST returnthe "unknownAnchor" error. A log is not able to generate an SCT for asubmission if itdoes not have access to the issuer's public key.¶
If the returnedsct
is intended to be provided to TLS clients, thensth
andinclusion
(if returned)SHOULD also be provided to TLS clients. Forexample, iftype
was 2 (indicatingprecert_sct_v2
), then all threeTransItem
s could beembedded in the certificate.¶
GET <Base URL>/ct/v2/get-sth¶
No inputs.¶
TransItem
of typesigned_tree_head_v2
signed by this log that is no older than the log's MMD.¶GET <Base URL>/ct/v2/get-sth-consistency¶
tree_size
of the older tree, in decimal.¶tree_size
of the newer tree, in decimal (optional).¶Both tree sizes must be from existing v2 STHs. However, because of skew, the receiving front end may not know one or both of the existing STHs. If both are known, then only theconsistency
output is returned. If the first is known but the second is not (or has been omitted), then the latest known STH is returned, along with a consistency proof between the first STH and the latest. If neither are known, then the latest known STH is returned without a consistency proof.¶
TransItem
of typeconsistency_proof_v2
whosetree_size_1
MUST match thefirst
input. If thesth
output is omitted, thentree_size_2
MUST match thesecond
input. Iffirst
andsecond
are equal and correspond to a known STH, the returned consistency proofMUST be empty (aconsistency_path
array with zero elements).¶TransItem
of typesigned_tree_head_v2
, signed by this log.¶Note that no signature is required for theconsistency
output, as it is used to verify the consistency between two signed STHs.¶
Error codes:¶
type | detail |
---|---|
firstUnknown | first is before the latest known STH but is not from an existing STH. |
secondUnknown | second is before the latest known STH but is not from an existing STH. |
secondBeforeFirst | second is smaller thanfirst . |
SeeSection 2.1.4.2 for an outline of how to use theconsistency
output.¶
GET <Base URL>/ct/v2/get-proof-by-hash¶
tree_size
of the tree on which to base the proof, in decimal.¶Thehash
must be calculated as defined inSection 4.7. A v2 STH must exist for thetree_size
. Because of skew, the front end may not know the requested tree head. In that case, it will return the latest STH it knows, along with an inclusion proof to that STH. If the front end knows the requested tree head, then onlyinclusion
is returned.¶
TransItem
of typeinclusion_proof_v2
whoseinclusion_path
array of Merkle Tree nodes proves the inclusion of the certificate (as specified by thehash
parameter) in the selected STH.¶TransItem
of typesigned_tree_head_v2
, signed by this log.¶Note that no signature is required for theinclusion
output, as it is used to verify inclusion in the selected STH, which is signed.¶
Error codes:¶
type | detail |
---|---|
hashUnknown | hash is not the hash of a known leaf (may be caused by skew or by a known certificate not yet merged). |
treeSizeUnknown | hash is before the latest known STH but is not from an existing STH. |
SeeSection 2.1.3.2 for an outline of how to use theinclusion
output.¶
GET <Base URL>/ct/v2/get-all-by-hash¶
tree_size
of the tree on which to base the proofs, in decimal.¶Thehash
must be calculated as defined inSection 4.7. A v2 STH must exist for thetree_size
.¶
Because of skew, the front end may not know the requested tree head or the requested hash, which leads to a number of cases:¶
Case | Response |
---|---|
latest STH < requested tree head | Return latest STH. |
latest STH > requested tree head | Return latest STH and a consistency proof between it and the requested tree head (seeSection 5.3). |
index of requested hash < latest STH | Returninclusion . |
Note that more than one case can be true; in which case, the returned data istheir union. It is also possible for none to be true; in which case, thefront endMUST return an empty response.¶
TransItem
of typeinclusion_proof_v2
whoseinclusion_path
array of Merkle Tree nodes proves the inclusion of the certificate (as specified by thehash
parameter) in the selected STH.¶TransItem
of typesigned_tree_head_v2
, signed by this log.¶TransItem
of typeconsistency_proof_v2
that proves the consistency of the requested tree head and the returned STH.¶Note that no signature is required for theinclusion
orconsistency
outputs, as they are used to verify inclusion in and consistency of signed STHs.¶
Errors are the same as inSection 5.4.¶
SeeSection 2.1.3.2 for an outline of how to use theinclusion
output,and seeSection 2.1.4.2 for an outline of how to use theconsistency
output.¶
GET <Base URL>/ct/v2/get-entries¶
An array of objects, each consisting of:¶
TransItem
structure of typex509_entry_v2
orprecert_entry_v2
(seeSection 4.3).¶submit-entry
, with the addition of the trust anchor to thechain
field if the submission did not include it.¶TransItem
of typex509_sct_v2
orprecert_sct_v2
, corresponding to this log entry.¶TransItem
of typesigned_tree_head_v2
, signed by this log.¶Note that this message is not signed -- theentries
data can be verified byconstructing the Merkle Tree Hash corresponding to a retrieved STH. All leavesMUST be v2. However, a compliant v2 clientMUST NOT construe an unrecognizedTransItem
type as an error. This means it may be unable to parse some entries,but note that each client can inspect the entries it does recognize as well asverify the integrity of the data by treating unrecognized leaves as opaque inputto the tree.¶
Thestart
andend
parametersSHOULD be within the range 0 <= x <tree_size
,as returned byget-sth
inSection 5.2.¶
Thestart
parameterMUST be less than or equal to theend
parameter.¶
Eachsubmitted_entry
output parameterMUST include the trust anchor that thelog used to verify thesubmission
, even if that trust anchor was not providedtosubmit-entry
(seeSection 5.1). If thesubmission
does not certifyitself, then the first element ofchain
MUST be present andMUST certify thesubmission
.¶
Log serversMUST honor requests where 0 <=start
<tree_size
andend
>=tree_size
by returning a partial response covering only the valid entries inthe specified range.end
>=tree_size
could be caused by skew. Note that thefollowing restriction may also apply:¶
LogsMAY restrict the number of entries that can be retrieved perget-entries
request. If a client requests more than the permitted number of entries, the logSHALL return the maximum number of entries permissible. These entriesSHALL besequential beginning with the entry specified bystart
.Note that a limit on the number of entries is not immutable, and thereforethe restriction may be changed or lifted at any time and is not listedwith the other Log Parameters inSection 4.1.¶
Because of skew, it is possible the log server will not have any entries betweenstart
andend
. In this case, itMUST return an emptyentries
array.¶
In any case, the log serverMUST return the latest STH it knows about.¶
SeeSection 2.1.2 for an outline of how to use a complete list oflog_entry
entries to verify theroot_hash
.¶
Error codes:¶
type | detail |
---|---|
startUnknown | start is greater than the number of entries in the Merkle Tree. |
endBeforeStart | start cannot be greater thanend . |
GET <Base URL>/ct/v2/get-anchors¶
No inputs.¶
This data is not signed, and the protocol depends on the security guarantees of TLS to ensure correctness.¶
CT-using TLS serversMUST use at least one of the mechanisms described belowto present one or more SCTs from one or more logs to each TLS client during fullTLS handshakes, when requested by the client, where each SCT corresponds to the server certificate.(Of course, a server can only send a TLS extension if the client hasspecified it first.)ServersSHOULD also present corresponding inclusion proofs and STHs.¶
A server can provide SCTs usinga TLS 1.3 extension (Section 4.2 of [RFC8446]) with typetransparency_info
(seeSection 6.5). This mechanism allows TLS servers toparticipate in CT without the cooperation of CAs, unlike the other twomechanisms. It also allows SCTs and inclusion proofs to be updated on the fly.¶
The server may also use an Online Certificate Status Protocol (OCSP)[RFC6960] response extension (seeSection 7.1.1),providing the OCSP response as part of the TLS handshake. Providinga response during a TLS handshake is popularly known as "OCSP stapling".For TLS1.3, the information is encoded as an extension in thestatus_request
extension data; seeSection 4.4.2.1 of [RFC8446]. For TLS 1.2[RFC5246], the informationis encoded in theCertificateStatus
message; seeSection 8 of [RFC6066]. Using stapling alsoallows SCTs and inclusion proofs to be updated on the fly.¶
CT information can also be encoded as an extension in the X.509v3 certificate(seeSection 7.1.2). Thismechanism allows the use of unmodified TLS servers, but the SCTs and inclusionproofs cannot be updated on the fly. Since the logs from which the SCTs andinclusion proofs originated won't necessarily be accepted by TLS clients forthe full lifetime of the certificate, there is a risk that TLS clients maysubsequently consider the certificate to be noncompliant. In such an event, one ofthe other two mechanisms will need to be used to deliver CT information, or, if this isnot possible, the certificate will need to be reissued.¶
This specification includes no description of how a TLS server canuse CT for TLS client certificates.While this may be useful, it is not documented here for the followingreasons:¶
A future version could include such information.¶
CT-using TLS serversSHOULD send SCTs from multiple logs because:¶
To select the logs from which to obtain SCTs, a TLS server can, for example,examine the set of logs popular TLS clients accept and recognize.¶
Multiple SCTs, inclusion proofs, and indeedTransItem
structures of any type are combined into a list as follows:¶
opaque SerializedTransItem<1..2^16-1>; struct { SerializedTransItem trans_item_list<1..2^16-1>; } TransItemList;¶
Here,SerializedTransItem
is an opaque byte string that contains the serializedTransItem
structure. This encoding ensures that TLS clients can decode eachTransItem
individually (so, for example, if there is a version upgrade, out-of-date clients can still parse oldTransItem
structures while skipping over newTransItem
structures whose versions they don't understand).¶
In eachTransItemList
that is sent during a TLS handshake, the TLSserverMUST include aTransItem
structure of typex509_sct_v2
orprecert_sct_v2
.¶
Presenting inclusion proofs and STHs in the TLS handshake helps to protect theclient's privacy (seeSection 8.1.4) and reduces load on logservers. Therefore, if the TLS server can obtain them, itSHOULD also includeTransItem
s of typeinclusion_proof_v2
andsigned_tree_head_v2
in theTransItemList
.¶
Provided that a TLS client includes thetransparency_info
extension type inthe ClientHello and the TLS server supports thetransparency_info
extension:¶
extension_data
is empty.¶TransItemList
of relevantTransItem
s (seeSection 6.4), whichSHOULD omit anyTransItem
s that are already embedded in the server certificate or the stapled OCSP response (seeSection 7.1). If the constructedTransItemList
is not empty, then the TLS serverMUST include thetransparency_info
extension with theextension_data
set to thisTransItemList
. If the list is empty, then the serverSHOULD omit theextension_data
element butMAY send it with an empty array.¶TLS serversMUST only include this extension in the following messages:¶
TLS serversMUST NOT process or include this extension when a TLS session isresumed, since session resumption uses the original session information.¶
The Transparency Information X.509v3 extension, which has OID 1.3.101.75 andSHOULD be noncritical, contains one or moreTransItem
structures in aTransItemList
. This extensionMAY be included in OCSP responses (seeSection 7.1.1) and certificates (seeSection 7.1.2). Since[RFC5280] requires theextnValue
field (an OCTET STRING) of each X.509v3 extension to include the DER encoding of an ASN.1 value, aTransItemList
MUST NOT be included directly. Instead, itMUST be wrapped inside an additional OCTET STRING, which is then put into theextnValue
field:¶
TransparencyInformationSyntax ::= OCTET STRING¶
TransparencyInformationSyntax
contains aTransItemList
.¶
A certification authorityMAY include a Transparency Information X.509v3 extension in thesingleExtensions
of aSingleResponse
in an OCSP response. All included SCTs and inclusion proofsMUST be for the certificate identified by thecertID
of thatSingleResponse
or for a precertificate that corresponds to that certificate.¶
A certification authorityMAY include a Transparency Information X.509v3extension in a certificate. All included SCTs and inclusion proofsMUST be for aprecertificate that corresponds to this certificate.¶
A certification authoritySHOULD NOT issue any certificate that identifies thetransparency_info
TLS extension in a TLS feature extension[RFC7633], becauseTLS servers are not required to support thetransparency_info
TLS extension inorder to participate in CT (seeSection 6).¶
There are various different functions clients of logs might perform. We describe here some typical clients and how they should function. Any inconsistency may be used as evidence that a log has not behaved correctly, and the signatures on the data structures prevent the log from denying that misbehavior.¶
All clients need various parameters in order to communicate with logs and verify their responses. These parameters are described inSection 4.1, but note that this document does not describe how the parameters are obtained, which is implementation dependent (for example, see[Chromium.Policy]).¶
TLS clients receive SCTs and inclusion proofs alongside or in certificates. CT-using TLS clientsMUST implement all of the three mechanisms by which TLS servers may present SCTs (seeSection 6).¶
TLS clients that support thetransparency_info
TLS extension (seeSection 6.5)SHOULD include it in ClientHello messages, with emptyextension_data
. If a TLS server includes thetransparency_info
TLS extension when resuming a TLS session, the TLS clientMUST abort the handshake.¶
Validation of an SCT for a certificate (where thetype
of theTransItem
isx509_sct_v2
) uses the unmodified TBSCertificate component of the certificate.¶
Before an SCT for a precertificate (where thetype
of theTransItem
isprecert_sct_v2
) can be validated, the TBSCertificate component of theprecertificate needs to be reconstructed from the TBSCertificate component ofthe certificate as follows:¶
In order to make use of a received SCT, the TLS clientMUST first validate it asfollows:¶
Compute the signature input by constructing aTransItem
of typex509_entry_v2
orprecert_entry_v2
, depending on the SCT'sTransItem
type. TheTimestampedCertificateEntryDataV2
structure is constructed in the following manner:¶
timestamp
is copied from the SCT.¶tbs_certificate
is the reconstructed TBSCertificate portion of the server certificate, as described inSection 8.1.2.¶issuer_key_hash
is computed as described inSection 4.7.¶sct_extensions
is copied from the SCT.¶signature
against the computed signature input using thepublic key of the corresponding log, which is identified by thelog_id
. Therequired signature algorithm is one of the log's parameters.¶If the TLS client does not have the corresponding log's parameters, it cannotattempt to validate the SCT. When evaluating compliance (seeSection 8.1.6), the TLS client will consider only those SCTs that itwas able to validate.¶
Note that SCT validation is not a substitute for the normal validation of theserver certificate and its chain.¶
When a TLS client has validated a received SCT but does not yet possessa corresponding inclusion proof, the TLS clientMAY request the inclusionproof directly from a log usingget-proof-by-hash
(Section 5.4) orget-all-by-hash
(Section 5.5).¶
Note that fetching inclusion proofs directly from a log will disclose to thelog which TLS server the client has been communicating with. This may beregarded as a significant privacy concern, and so it is preferable for the TLSserver to send the inclusion proofs (seeSection 6.4).¶
When a TLS client has received, or fetched, an inclusion proof (and an STH),itSHOULD proceed to verify the inclusion proof to the provided STH.The TLS clientSHOULD also verify consistency between the provided STHand an STH it knows about.¶
If the TLS client holds an STH that predates the SCT, itMAY, in the process ofauditing, request a new STH from the log (Section 5.2) and then verify it byrequesting a consistency proof (Section 5.3). Note that if the TLSclient usesget-all-by-hash
, then it will already have the new STH.¶
It is up to a client's local policy to specify the quantity and form ofevidence (SCTs, inclusion proofs, or a combination) needed to achievecompliance and how to handle noncompliance.¶
A TLS client can only evaluate compliance if it has given the TLS server theopportunity to send SCTs and inclusion proofs by any of the three mechanismsthat are mandatory to implement for CT-using TLS clients (seeSection 8.1.1). Therefore, a TLS clientMUST NOT evaluate complianceif it did not include both thetransparency_info
andstatus_request
TLSextensions in the ClientHello.¶
Monitors watch logs to check for correct behavior, for certificates ofinterest, or for both. For example, a monitor may be configured to report on allcertificates that apply to a specific domain name when fetching new entries forconsistency validation.¶
A monitorMUST at least inspect every new entry in every log it watches, and itMAY also choose to keep copies of entire logs.¶
To inspect all of the existing entries, the monitorSHOULD follow these stepsonce for each log:¶
To inspect new entries, the monitorSHOULD follow these steps repeatedly for each log:¶
Either:¶
Or, if it is not keeping all log entries:¶
Auditing ensures that the current published state of a log is reachable frompreviously published states that are known to be good and that the promisesmade by the log, in the form of SCTs, have been kept. Audits are performed bymonitors or TLS clients.¶
In particular, there are four properties of log behavior that should be checked:¶
A benign, conformant log publishes a series of STHs over time, each derived fromthe previous STH and the submitted entries incorporated into the log sincepublication of the previous STH. This can be proven through auditing of STHs.SCTs returned to TLS clients can be audited by verifying against theaccompanying certificate and using Merkle inclusion proofs against the log'sMerkle Tree.¶
The action taken by the auditor, if an audit fails, is not specified, but notethat in general, if an audit fails, the auditor is in possession of signed proof ofthe log's misbehavior.¶
A monitor (Section 8.2) can audit by verifying the consistency of STHs itreceives, ensuring that each entry can be fetched and that the STH is indeed theresult of making a tree from all fetched entries.¶
A TLS client (Section 8.1) can audit by verifying an SCT against any STHdated after the SCT timestamp + the Maximum Merge Delay by requesting a Merkleinclusion proof (Section 5.4). It can also verify that the SCTcorresponds to the server certificate it arrived with (i.e., the log entry isthat certificate or is a precertificate corresponding to that certificate).¶
Checking of the consistency of the log view presented to all entities is moredifficult to perform because it requires a way to share log responses among aset of CT-using entities and is discussed inSection 11.3.¶
It is not possible for a log to change either of its algorithms part way throughits lifetime:¶
Allowing multiple signature or hash algorithms for a log would require that alldata structures support it and would significantly complicate clientimplementation, which is why it is not supported by this document.¶
If it should become necessary to deprecate an algorithm used by a live log, thenthe logMUST be frozen, as specified inSection 4.13, and a new logSHOULD bestarted. Certificates in the frozen log that have not yet expired and requirenew SCTsSHOULD be submitted to the new log and the SCTs from that log usedinstead.¶
The assignment policy criteria mentioned in this section refer to the policiesoutlined in[RFC8126].¶
This subsection defines additions to existing registries.¶
IANA has added the following entryto the "TLS ExtensionType Values" registry defined in[RFC8446],with an assigned Value:¶
Value | Extension Name | TLS 1.3 | DTLS-Only | Recommended | Reference |
---|---|---|---|---|---|
52 | transparency_info | CH, CR, CT | N | Y | RFC 9162 |
IANA has added a new entry in the "IETF URN Sub-namespace for Registered Protocol Parameter Identifiers" registry, following the template in[RFC3553]:¶
IANA has added a new protocol registry, "Public NotaryTransparency", to the list that appears at<https://www.iana.org/assignments/>¶
The rest of this section defines the subregistries that have been created within the new "Public Notary Transparency" registry.¶
IANA has established a registry of hash algorithm values, named"Hash Algorithms", with the following registration procedures:¶
Range | Registration Procedures |
---|---|
0x00-0xDF | Specification Required |
0xE0-0xEF | Experimental Use |
0xF0-0xFF | Private Use |
The "Hash Algorithms" registry initially consists of:¶
Value | Hash Algorithm | OID | Reference |
---|---|---|---|
0x00 | SHA-256 | 2.16.840.1.101.3.4.2.1 | [RFC6234] |
0x01 - 0xDF | Unassigned | RFC 9162 | |
0xE0 - 0xEF | Reserved for Experimental Use | RFC 9162 | |
0xF0 - 0xFF | Reserved for Private Use | RFC 9162 |
The designated expert(s) should ensure that the proposed algorithm has a publicspecification and is suitable for use as a cryptographic hash algorithm with noknown preimage or collision attacks. These attacks can damage the integrity ofthe log.¶
IANA has established a registry of signature algorithm values, named"Signature Algorithms".¶
The following notes have been added to the registry:¶
- Note:
- This is a subset of the "TLS SignatureScheme" registry, limited to those algorithms that are appropriate for CT. A major advantage of this is leveraging the expertise of the TLS Working Group and its designated expert(s).¶
- Note:
- The value
0x0403
appears twice. While this may be confusing, it is okay because the verification process is the same for both algorithms, and the choice of which to use when generating a signature is purely internal to the log server.¶
The "Signature Algorithms" registry has the following registration procedures:¶
Range | Registration Procedures |
---|---|
0x0000-0x0807 | Specification Required |
0x0808-0xFDFF | Expert Review |
0xFE00-0xFEFF | Experimental Use |
0xFF00-0xFFFF | Private Use |
The "Signature Algorithms" registry initially consists of:¶
SignatureScheme Value | Signature Algorithm | Reference |
---|---|---|
0x0000 - 0x0402 | Unassigned | |
ecdsa_secp256r1_sha256 (0x0403) | ECDSA (NIST P-256) with SHA-256 | [FIPS186-4] |
ecdsa_secp256r1_sha256 (0x0403) | Deterministic ECDSA (NIST P-256) with HMAC-SHA256 | [RFC6979] |
0x0404 - 0x0806 | Unassigned | |
ed25519 (0x0807) | Ed25519 (PureEdDSA with the edwards25519 curve) | [RFC8032] |
0x0808 - 0xFDFF | Unassigned | |
0xFE00 - 0xFEFF | Reserved for Experimental Use | RFC 9162 |
0xFF00 - 0xFFFF | Reserved for Private Use | RFC 9162 |
The designated expert(s) should ensure that the proposed algorithm has a publicspecification, has a value assigned to it in the "TLS SignatureScheme" registry(which was established by[RFC8446]), and is suitable for use as acryptographic signature algorithm.¶
IANA has established a registry ofVersionedTransType
values, named"VersionedTransTypes".¶
The following note has been added:¶
- Note:
- The range 0x0000..0x00FF is reserved so that v1 SCTs are distinguishable from v2 SCTs and other
TransItem
structures.¶
The registration procedures for the "VersionedTransTypes" registry are the following:¶
Range | Registration Procedures |
---|---|
0x0100-0xDFFF | Specification Required |
0xE000-0xEFFF | Experimental Use |
0xF000-0xFFFF | Private Use |
The "VersionedTransTypes" registry initially consists of:¶
Value | Type and Version | Reference |
---|---|---|
0x0000 - 0x00FF | Reserved | [RFC6962] |
0x0100 | x509_entry_v2 | RFC 9162 |
0x0101 | precert_entry_v2 | RFC 9162 |
0x0102 | x509_sct_v2 | RFC 9162 |
0x0103 | precert_sct_v2 | RFC 9162 |
0x0104 | signed_tree_head_v2 | RFC 9162 |
0x0105 | consistency_proof_v2 | RFC 9162 |
0x0106 | inclusion_proof_v2 | RFC 9162 |
0x0107 - 0xDFFF | Unassigned | |
0xE000 - 0xEFFF | Reserved for Experimental Use | RFC 9162 |
0xF000 - 0xFFFF | Reserved for Private Use | RFC 9162 |
The designated expert(s) should review the public specification to ensure that it isdetailed enough to ensure implementation interoperability.¶
IANA has established a registry ofExtensionType
values, named "LogArtifact Extensions".¶
The registration procedures for the "Log Artifact Extensions" registry are the following:¶
Range | Registration Procedures |
---|---|
0x0000-0xDFFF | Specification Required |
0xE000-0xEFFF | Experimental Use |
0xF000-0xFFFF | Private Use |
The "Log Artifact Extensions" registry initially consists of:¶
ExtensionType | Status | Use | Reference |
---|---|---|---|
0x0000 - 0xDFFF | Unassigned | n/a | |
0xE000 - 0xEFFF | Reserved for Experimental Use | n/a | RFC 9162 |
0xF000 - 0xFFFF | Reserved for Private Use | n/a | RFC 9162 |
The "Use" column should contain one or both of the following values:¶
The designated expert(s) should review the public specification to ensure that it isdetailed enough to ensure implementation interoperability. They shouldalso verify that the extension is appropriate to the contexts in which it isspecified to be used (SCT, STH, or both).¶
IANA has established a registry of Log IDs, named "Log IDs".¶
The registry's registration procedure is First Come First Served.¶
The "Log IDs" registry initially consists of:¶
Log ID | Log Base URL | Log Operator | Reference |
---|---|---|---|
1.3.101.8192 - 1.3.101.16383 | Unassigned | Unassigned | |
1.3.101.80.0 - 1.3.101.80.* | Unassigned | Unassigned |
The following notes have been added to the registry:¶
- Note:
- All OIDs in the range from 1.3.101.8192 to 1.3.101.16383 have been set asidefor Log IDs.This is a limited resource of 8,192 OIDs, each of which has an encoded length of4 octets.¶
- Note:
- The 1.3.101.80 arc has also been set aside for Log IDs.This is an unlimited resource, but onlythe 128 OIDs from 1.3.101.80.0 to 1.3.101.80.127 have an encoded length of only4 octets.¶
Each application for the allocation of a Log IDMUST be accompanied by:¶
IANA is asked to reject any request to update a Log ID or Log Base URL in thisregistry because these fields are immutable (seeSection 4.1).¶
IANA is asked to accept requests from log operators to update their contactdetails in this registry.¶
Since log operators can choose to not use this registry (seeSection 4.4), it isnot expected to be a global directory of all logs.¶
IANA has created a new registry for errors,the "Error Types" registry.¶
The registration procedure for this registry is Specification Required.¶
This registry has the following three fields:¶
Field Name | Type | Reference |
---|---|---|
Identifier | string | RFC 9162 |
Meaning | string | RFC 9162 |
Reference | string | RFC 9162 |
The initial values of the "Error Types" registry, which are taken from the text inSection 5, are as follows:¶
Identifier | Meaning | Reference |
---|---|---|
malformed | The request could not be parsed. | RFC 9162 |
badSubmission | submission is neither a valid certificate nor a valid precertificate. | RFC 9162 |
badType | type is neither 1 nor 2. | RFC 9162 |
badChain | The first element ofchain is not the certifier of thesubmission , or the second element does not certify the first, etc. | RFC 9162 |
badCertificate | One or more certificates inchain are not valid (e.g., not properly encoded). | RFC 9162 |
unknownAnchor | The last element ofchain (or, ifchain is an empty array, thesubmission ) is not, nor is it certified by, an accepted trust anchor. | RFC 9162 |
shutdown | The log is no longer accepting submissions. | RFC 9162 |
firstUnknown | first is before the latest known STH but is not from an existing STH. | RFC 9162 |
secondUnknown | second is before the latest known STH but is not from an existing STH. | RFC 9162 |
secondBeforeFirst | second is smaller thanfirst . | RFC 9162 |
hashUnknown | hash is not the hash of a known leaf (may be caused by skew or by a known certificate not yet merged). | RFC 9162 |
treeSizeUnknown | hash is before the latest known STH but is not from an existing STH. | RFC 9162 |
startUnknown | start is greater than the number of entries in the Merkle Tree. | RFC 9162 |
endBeforeStart | start cannot be greater thanend . | RFC 9162 |
IANA has assigned an object identifier from the "SMISecurity for PKIX Module Identifier" registry to identify theASN.1 module inAppendix B of this document.¶
Decimal | Description | References |
---|---|---|
102 | id-mod-public-notary-v2 | RFC 9162 |
With CAs, logs, and servers performing the actions described here, TLS clientscan use logs and signed timestamps to reduce the likelihood that they willaccept misissued certificates. If a server presents a valid signed timestamp fora certificate, then the client knows that a log has committed to publishing thecertificate. From this, the client knows that monitors acting for the subject ofthe certificate have had some time to notice the misissuance and take someaction, such as asking a CA to revoke a misissued certificate. A signedtimestamp does not guarantee this, though, since appropriate monitors might nothave checked the logs or the CA might have refused to revoke the certificate.¶
In addition, if TLS clients will not accept unlogged certificates, then siteowners will have a greater incentive to submit certificates to logs, possiblywith the assistance of their CA, increasing the overall transparency of thesystem.¶
Misissued certificates that have not been publicly logged, and thus do not havea valid SCT, are not considered compliant. Misissued certificates that do havean SCT from a log will appear in that public log within the Maximum Merge Delay,assuming the log is operating correctly. Since a log is allowed to serve an STHof any age up to the MMD, the maximum period of time during which a misissuedcertificate can be used without being available for audit is twice the MMD.¶
The logs do not themselves detect misissued certificates; they rely instead oninterested parties, such as domain owners, to monitor them and take correctiveaction when a misissue is detected.¶
A log can misbehave in several ways. Examples include the following: failing to incorporate acertificate with an SCT in the Merkle Tree within the MMD; presenting different,conflicting views of the Merkle Tree at different times and/or to differentparties; issuing STHs too frequently; mutating the signature of a loggedcertificate; and failing to present a chain containing the certifier of a loggedcertificate.¶
Violation of the MMD contract is detected by log clients requesting a Merkleinclusion proof (Section 5.4) for each observed SCT. These checks canbe asynchronous and need only be done once per certificate. However, note thatthere may be privacy concerns (seeSection 8.1.4).¶
Violation of the append-only property or the STH issuance rate limit can be detected by multiple clients comparing their instances of the STHs. This technique, known as "gossip", is an active area of research and not defined here. Proof of misbehavior in such cases would be either a series of STHs that were issued too closely together, proving violation of the STH issuance rate limit, or an STH with a root hash that does not match the one calculated from a copy of the log, proving violation of the append-only property.¶
Clients that report back SCTs can be tracked or traced if a logproduces multiple STHs or SCTs with the same timestamp and data but differentsignatures. LogsSHOULD mitigate this risk by either:¶
By requiring TLS servers to offer multiple SCTs, each from a different log, TLSclients reduce the effectiveness of an attack where a CA and a log collude(seeSection 6.2).¶
Malicious monitors can use logs to learn about the existence of domain namesthat might not otherwise be easy to discover. Some subdomain labels may revealinformation about the service and software for which the subdomain is used,which in turn might facilitate targeted attacks.¶
Certificate Transparency logs have to be either v1 (conforming to[RFC6962]) orv2 (conforming to this document), as the data structures are incompatible, and soa v2 log could not issue a valid v1 SCT.¶
CT clients, however, can support v1 and v2 SCTs for the same certificatesimultaneously, as v1 SCTs are delivered in different TLS, X.509, and OCSPextensions than v2 SCTs.¶
v1 and v2 SCTs for X.509 certificates can be validated independently. Forprecertificates, v2 SCTs should be embedded in the TBSCertificate beforesubmission of the TBSCertificate (inside a v1 precertificate, as described inSection 3.1 of [RFC6962]) to a v1 log so that TLS clients conforming to[RFC6962] but not this document are oblivious to the embedded v2 SCTs. An issuercan follow these steps to produce an X.509 certificate with embedded v1 and v2SCTs:¶
The following ASN.1[X.680] module may be useful to implementors. This module references[RFC5912] and[RFC6268].¶
CertificateTransparencyV2Module-2021 -- { id-mod-public-notary-v2 from above, in iso(1) identified-organization(3) ... form }DEFINITIONS IMPLICIT TAGS ::= BEGIN-- EXPORTS ALL --IMPORTS EXTENSION FROM PKIX-CommonTypes-2009 -- RFC 5912 { iso(1) identified-organization(3) dod(6) internet(1) security(5) mechanisms(5) pkix(7) id-mod(0) id-mod-pkixCommon-02(57) } CONTENT-TYPE FROM CryptographicMessageSyntax-2010 -- RFC 6268 { iso(1) member-body(2) us(840) rsadsi(113549) pkcs(1) pkcs-9(9) smime(16) modules(0) id-mod-cms-2009(58) } TBSCertificate FROM PKIX1Explicit-2009 -- RFC 5912 { iso(1) identified-organization(3) dod(6) internet(1) security(5) mechanisms(5) pkix(7) id-mod(0) id-mod-pkix1-explicit-02(51) };---- Section 3.2. Precertificates--ct-tbsCertificate CONTENT-TYPE ::= { TYPE TBSCertificate IDENTIFIED BY id-ct-tbsCertificate }id-ct-tbsCertificate OBJECT IDENTIFIER ::= { 1 3 101 78 }---- Section 7.1. Transparency Information X.509v3 Extension--ext-transparencyInfo EXTENSION ::= { SYNTAX TransparencyInformationSyntax IDENTIFIED BY id-ce-transparencyInfo CRITICALITY { FALSE } }id-ce-transparencyInfo OBJECT IDENTIFIER ::= { 1 3 101 75 }TransparencyInformationSyntax ::= OCTET STRING---- Section 7.1.1. OCSP Response Extension--ext-ocsp-transparencyInfo EXTENSION ::= { SYNTAX TransparencyInformationSyntax IDENTIFIED BY id-pkix-ocsp-transparencyInfo CRITICALITY { FALSE } }id-pkix-ocsp-transparencyInfo OBJECT IDENTIFIER ::= id-ce-transparencyInfo---- Section 8.1.2. Reconstructing the TBSCertificate--ext-embeddedSCT-CTv1 EXTENSION ::= { SYNTAX SignedCertificateTimestampList IDENTIFIED BY id-ce-embeddedSCT-CTv1 CRITICALITY { FALSE } }id-ce-embeddedSCT-CTv1 OBJECT IDENTIFIER ::= { 1 3 6 1 4 1 11129 2 4 2 }SignedCertificateTimestampList ::= OCTET STRINGEND¶
The authors would like to thankErwann Abelea,Robin Alden,Andrew Ayer,Richard Barnes,Al Cutter,David Drysdale,Francis Dupont,Adam Eijdenberg,Stephen Farrell,Daniel Kahn Gillmor,Paul Hadfield,Brad Hill,Jeff Hodges,Paul Hoffman,Jeffrey Hutzelman,Kat Joyce,Emilia Kasper,Stephen Kent,Adam Langley,SM,Alexey Melnikov,Linus Nordberg,Chris Palmer,Trevor Perrin,Pierre Phaneuf,Eric Rescorla,Rich Salz,Melinda Shore,Ryan Sleevi,Martin Smith,Carl Wallace, andPaul Wouters for their valuable contributions.¶
A big thank you to Symantec for kindly donating the OIDs from the 1.3.101 arc that are used in this document.¶