Hi,
I reviewed this document as part of the IoT Directorate's ongoing effort to
review all IETF documents being processed by the IESG. These comments were
written primarily for the benefit of the Security Area Directors. Document
authors, document editors, and WG chairs should treat these comments just like
any other IETF Last Call comments.
Review Results: Ready with Nits
Please find my comments below.
Yours,
Daniel
Deprecating MD5 and SHA-1 signature hashes in TLS 1.2
draft-ietf-tls-md5-sha1-deprecate-04
[...]
1. Introduction
The usage of MD5 and SHA-1 for signature hashing in TLS 1.2 is
specified in [RFC5246]. MD5 and SHA-1 have been proven to be
insecure, subject to collision attacks [Wang]. In 2011, [RFC6151]
detailed the security considerations, including collision attacks for
MD5. NIST formally deprecated use of SHA-1 in 2011
[NISTSP800-131A-R2] and disallowed its use for digital signatures at
the end of 2013, based on both the Wang, et. al, attack and the
potential for brute-force attack. In 2016, researchers from INRIA
identified a new class of transcript collision attacks on TLS (and
other protocols) that rely on efficient collision-finding algorithms
on the underlying hash constructions [Transcript-Collision].
Further, in 2017, researchers from Google and CWI Amsterdam
[SHA-1-Collision] proved SHA-1 collision attacks were practical.
This document updates [RFC5246] and [RFC7525] in such a way that MD5
and SHA-1 MUST NOT be used for digital signatures. However, this
document does not deprecate SHA-1 in HMAC for record protection.
RFC6194 may be mentioned as a reference for
not deprecating HMAC-SHA-1 as well as an
additional reference to [NISTSP800-131A-R2].
Reading the text the situation of HMAC with
MD5 is unclear. Since we specify that SHA-1
is not deprecated for HMAC we may specify
the status for HMAC with MD5. Given RFC6151 I
hope the reason is that MD5 and HMAC-MD5 has
already been deprecated but I have not found
this. Maybe that would worth mentioning it
is deprecated already.
[...]
2. Signature Algorithms
Clients MUST NOT include MD5 and SHA-1 in the signature_algorithms
extension. If a client does not send a signature_algorithms
extension, then the server MUST abort the handshake and send a
handshake_failure alert, except when digital signatures are not used
(for example, when using PSK ciphers).
It seems to me that the server behavior might
be defined as well. In our case this could be
something around the lines the server MUST
ignore MD5 and SHA1 values in the signature
algorithm extension.
3. Certificate Request
Servers SHOULD NOT include MD5 and SHA-1 in CertificateRequest
messages.
It seems to me that the same level of
authentication should be provided for both
peers and that server MUST NOT include MD5
or SHA-1.
A SHOULD NOT status might be welcome for a
smooth transition. At that time, collision
for MD5 and SHA1 are known for years. It is likely
that software that still need MD5 or SHA1 are
likely to never upgrade, so I doubt a smooth
path worth being taken.
4. Server Key Exchange
Servers MUST NOT include MD5 and SHA-1 in ServerKeyExchange messages.
If a client receives a MD5 or SHA-1 signature in a ServerKeyExchange
message it MUST abort the connection with the illegal_parameter
alert.
As per section 2, the client has clearly
indicated it does not support signature with
MD5/SHA1, so Server Key Exchange should not
end up with signature with SHA1/MD5.
"""
If the client has offered the "signature_algorithms" extension, the
signature algorithm and hash algorithm MUST be a pair listed in that
extension.
"""
It also seems to me that the constraint of
including a MD5 and SHA-1 signature is
related to the Certificate. I suspect that
some clarification are needed here.
Since the case where the extension becomes
mandatory, the quoted text above of RFC 5246
might be updated as well, though this does
not appear that necessary.
5. Certificate Verify
Clients MUST NOT include MD5 and SHA-1 in CertificateVerify messages.
If a server receives a CertificateVerify message with MD5 or SHA-1 it
MUST abort the connection with handshake_failure or
insufficient_security alert.
6. Certificate
Unless I am missing something, it seems to me
that signature may also be found in the
Certificate messages for the chain as well in
the restriction of the signature algorithm.
The end certificate is associated to the peer
while other certificate are related to a CA.
It seems that client and server behavior may
be specified. The quoted text below may be
helpful to clarify.
"""
If the client provided a "signature_algorithms" extension, then all
certificates provided by the server MUST be signed by a
hash/signature algorithm pair that appears in that extension.
"""
6. Updates to RFC5246
[RFC5246], The Transport Layer Security (TLS) Protocol Version 1.2,
suggests that implementations can assume support for MD5 and SHA-1 by
their peer. This update changes the suggestion to assume support for
SHA-256 instead, due to MD5 and SHA-1 being deprecated.
In Section 7.4.1.4.1: the text should be revised from:
OLD:
"Note: this is a change from TLS 1.1 where there are no explicit
rules, but as a practical matter one can assume that the peer
supports MD5 and SHA- 1."
NEW:
"Note: This is a change from TLS 1.1 where there are no explicit
rules, but as a practical matter one can assume that the peer
supports SHA-256."
I am reading the Note as an explanation on
why sha was taken as the default hash
function with the following rules.
"""
If the client does not send the signature_algorithms extension, the
server MUST do the following:
- If the negotiated key exchange algorithm is one of (RSA, DHE_RSA,
DH_RSA, RSA_PSK, ECDH_RSA, ECDHE_RSA), behave as if client had
sent the value {sha1,rsa}.
- If the negotiated key exchange algorithm is one of (DHE_DSS,
DH_DSS), behave as if the client had sent the value {sha1,dsa}.
- If the negotiated key exchange algorithm is one of (ECDH_ECDSA,
ECDHE_ECDSA), behave as if the client had sent value {sha1,ecdsa}.
"""
The current document does not update the
default hash function from sha to sha256 to
avoid interoperability issue where one peer
takes sha while the other one takes sha-256.
As a results, these rules and the "Note" may
eventually all together be replaced by your
text of section 2.
The following text may also be removed:
"""
If the client supports only the default hash and signature algorithms
(listed in this section), it MAY omit the signature_algorithms
extension.
"""
Regarding the Note, it seems to be that the
removal of support for MD5/SHA1 will result
in interoperability issues. At this point,
the issue is due to the obsolescence of the
implementation as deprecation of SHA1/Md5 has
started a long time ago.
It is unclear to me how normative is
interpreted "can assume". Was the support of
MD5/SHA1 a SHOULD or a MUST? In both case, if
we were willing to maintain interoperability
between software that only implemented
MD5/SHA1, we should take a slower path and
introducing SHA-256 and having were MD5/SHA1
kept for interoperability purpose before
being deprecated. I do not think we should
take that path as implementations that
currently do not support SHA-256 are unlikely
to be updated and that deprecation of
SHA1/MD5 has started a long time ago.
I would however mention the issue of
interoperability in the section but not in
the text to update. In the text to update I
would maybe suggest that the support of
SHA-256 comes with a normative MUST
statement.
Velvindron, et al. Expires April 12, 2021 [Page 3]
Internet-Draft draft-ietf-tls-md5-sha1-deprecate October 2020
7. Updates to RFC7525
[RFC7525], Recommendations for Secure Use of Transport Layer Security
(TLS) and Datagram Transport Layer Security (DTLS) recommends use of
SHA-256 as a minimum requirement. This update moves the minimum
recommendation to use stronger language deprecating use of both SHA-1
and MD5. The prior text did not explicitly include MD5 or SHA-1; and
this text adds guidance to ensure that these algorithms have been
deprecated..
Section 4.3:
OLD:
When using RSA, servers SHOULD authenticate using certificates with
at least a 2048-bit modulus for the public key. In addition, the use
of the SHA-256 hash algorithm is RECOMMENDED (see [CAB-Baseline] for
more details). Clients SHOULD indicate to servers that they request
SHA-256, by using the "Signature Algorithms" extension defined in TLS
1.2.
NEW:
Servers SHOULD authenticate using certificates with at least a
2048-bit modulus for the public key.
In addition, the use of the SHA-256 hash algorithm is RECOMMENDED;
and SHA-1 or MD5 MUST NOT be used (see [CAB-Baseline] for more
details). Clients MUST indicate to servers that they request SHA-
256, by using the "Signature Algorithms" extension defined in TLS
1.2.
I understand the reason we do specify that
hash algorithms that MUST NOT been used. This
is fine in the context of this document, but
it seems to me that if we were writing the
updated specification we may have rather
mentioned a minimum level of security hash
function needs to be met - in our case
SHA-256. I leave the co-authors make the
appropriated choice.
8. IANA Considerations
The document updates the "TLS SignatureScheme" registry to change the
recommended status of SHA-1 based signature schemes to N (not
recommended) as defined by [RFC8447]. The following entries are to
be updated:
+--------+----------------+-------------+-------------------+
| Value | Description | Recommended | Reference |
+--------+----------------+-------------+-------------------+
| 0x0201 | rsa_pkcs1_sha1 | N | [RFC8446][RFCTBD] |
| 0x0203 | ecdsa_sha1 | N | [RFC8446][RFCTBD] |
+--------+----------------+-------------+-------------------+
Other entries of the resgistry remain the same.
It seems to me that TLS 1.2 is using the TLS
hash and TLS signature registry TLS signature
registry and TLS 1.3 is using Signature
Scheme.
I suspect that TLS hash values for sha1 and
md5 should be deprecated. And RFCTBD should
be added for sha1 and md5. Note that the
SHOULD NOT status for CertificateRequest
may have prevented such deprecation.
A side effect is these code points for
signature scheme that were assigned for
compatibility with legacy (TLS 1.2)
signatures must not be used anymore - if
there are no more valid with TLS 1.2.