Re: Issue 195: Certificate Management API

I think these are different hash algorithms. There is a hash algorithm that
is used to sign the certificate and there is a hash algorithm that if used
to generate the fingerprint. These two, as far as I know, are independent.
This is why algorithm used to generate fingerprint is signaled in SDP.

_____________
Roman Shpount

On Tue, Sep 22, 2015 at 3:39 PM, Bernard Aboba <Bernard.Aboba@microsoft.com>
wrote:

> It appears that AlgorithmIdentifier does in some cases identify a complete
> ciphersuite, including both key generation and hash algorithms, and that
> some (*but not all*) the ciphersuites recommended by NIST are included.
>
> For example, here is what Section 18.5.2 of WebCrypto (
> http://www.w3.org/TR/WebCryptoAPI/#algorithms) says:
>
> *18.5.2. For Implementers** ¶
> <http://www.w3.org/TR/WebCryptoAPI/#algorithm-recommendations-implementers>*
>
> In order to promote interoperability for developers, this specification
> includes a list of suggested algorithms. These are considered to be the
> most widely used algorithms in practice at the time of writing, and
> therefore provide a good starting point for initial implementations of this
> specification. The suggested algorithms are:
>
>    - HMAC <http://www.w3.org/TR/WebCryptoAPI/#hmac> using SHA-1
>    <http://www.w3.org/TR/WebCryptoAPI/#alg-sha-1>
>    - HMAC <http://www.w3.org/TR/WebCryptoAPI/#hmac> using SHA-256
>    <http://www.w3.org/TR/WebCryptoAPI/#alg-sha-256>
>    - RSASSA-PKCS1-v1_5 <http://www.w3.org/TR/WebCryptoAPI/#rsassa-pkcs1>
>    using SHA-1 <http://www.w3.org/TR/WebCryptoAPI/#alg-sha-256>
>    - RSA-PSS <http://www.w3.org/TR/WebCryptoAPI/#rsa-pss> using SHA-256
>    <http://www.w3.org/TR/WebCryptoAPI/#alg-sha-256> and MGF1 with SHA-256
>    <http://www.w3.org/TR/WebCryptoAPI/#alg-sha-256>.
>    - RSA-OAEP <http://www.w3.org/TR/WebCryptoAPI/#rsa-oaep> using SHA-256
>    <http://www.w3.org/TR/WebCryptoAPI/#alg-sha-256> and MGF1 with SHA-256
>    <http://www.w3.org/TR/WebCryptoAPI/#alg-sha-256>.
>    - ECDSA <http://www.w3.org/TR/WebCryptoAPI/#ecdsa> using P-256
>    <http://www.w3.org/TR/WebCryptoAPI/#dfn-NamedCurve-p256> curve and
>    SHA-256 <http://www.w3.org/TR/WebCryptoAPI/#alg-sha-256>
>    - AES-CBC <http://www.w3.org/TR/WebCryptoAPI/#aes-cbc>
>
>
>
> Since we are talking about DTLS certificates, NIST Special Publication
> 800-52 (see:
> http://nvlpubs.nist.gov/nistpubs/SpecialPublications/NIST.SP.800-52r1.pdf)
> is relevant.  Table 3-1 includes the following recommendations:
>
>
>
>
>
> *From:* Roman Shpount [mailto:rshpount@turbobridge.com]
> *Sent:* Tuesday, September 22, 2015 10:24 AM
> *To:* Bernard Aboba <Bernard.Aboba@microsoft.com>
> *Cc:* public-ortc@w3.org
> *Subject:* Re: Issue 195: Certificate Management API
>
>
>
> Should we add optional parameter of type HashAlgorithmIdentifier (
> https://dvcs.w3.org/hg/webcrypto-api/raw-file/tip/spec/Overview.html#dfn-HashAlgorithmIdentifier
> <https://na01.safelinks.protection.outlook.com/?url=https%3a%2f%2fdvcs.w3.org%2fhg%2fwebcrypto-api%2fraw-file%2ftip%2fspec%2fOverview.html%23dfn-HashAlgorithmIdentifier&data=01%7c01%7cBernard.Aboba%40microsoft.com%7c649571601c584caa6e2708d2c3729556%7c72f988bf86f141af91ab2d7cd011db47%7c1&sdata=O0VcxxWePB7a1fKRt%2fFCi2yTtWDsNwa8QSpMKD2yhlI%3d>)
> to generateCertificate?
>
>
>
> It seems strange that it is not there already.
>
>
> _____________
> Roman Shpount
>
>
>
> On Tue, Sep 22, 2015 at 1:00 PM, Bernard Aboba <
> Bernard.Aboba@microsoft.com> wrote:
>
> WebRTC 1.0 added an (optional) Certificate Management API in Section 5.5
> in the last Editor’s draft.   This appears needed in ORTC API in order to
> solve the DtlsTransport forking problem.
>
>
>
> So below find the proposed text to be inserted in Section 15.   Some
> oddities found in the API:
>
>
>
> a.       keygenAlgorithm only specifies the algorithm used to generate
> the key – it does not specify the hash algorithm.  So if in future we
> needed to transition from SHA-256 to some other hash algorithm, it isn’t
> clear how that would happen.
>
> b.      In practice, implementations are likely to use the certificate
> generation functionality in the underlying DTLS implementation (e.g.
> BoringSSL, OpenSSL, S-Channel, etc.).  So why the references to WebCrypto?
>
>
>
>