[cabfpub] Ballot 108: Clarifying the scope of the baseline requirements

Ryan Sleevi sleevi at google.com
Fri Aug 2 00:46:50 UTC 2013


Moudrick,

The importance of clarifying that the intermediates are in scope is
that such intermediates are capable of issuing any kind of certificate
(including server certificates), so it's important to consider them in
scope.

The comment about permitting EKUs in intermediate certificates - and
their value - has been discussed in the context of Ballot 105. The
short answer is that several client libraries *further* restrict the
acceptable EKUs based on the transitive EKUs of the certificate chain
it appears in.

That is, if a subordinate CA certificate *has* an EKU extension, and
it lacks the aforementioned EKUs, then any certificates issued by that
CA will *not* be accepted for server auth - regardless of what EKUs
the leaf says. That is, treating EKUs in intermediates as "eku
constraints", for lack of a better term.

While this may be seen as controversial to "PKIX purists", it's been
deployed for over a decade, and offers a reasonable path for scoping
subordinate CAs. Note that RFC 5280 does not *prevent* this extension
from appearing in the intermediates, so such behaviour SHOULD be
acceptable to other software that does not implement this check (eg:
non-browser users of root stores)

Hopefully that clears things up.

Cheers,
Ryan

On Thu, Aug 1, 2013 at 3:56 PM, Moudrick M. Dadashov <md at ssc.lt> wrote:
> Jeremy, my comment was about Key Usage not EKU, sorry for the confuse..
>
> If the proposed change requires EKU for non EE certificates, then there is
> one more issue:
>
> RFC 5280:
> 4.2.1.12. Extended Key Usage
>
>    This extension indicates one or more purposes for which the certified
>    public key may be used, in addition to or in place of the basic
>    purposes indicated in the key usage extension.  In general, this
>    extension will appear only in end entity certificates.
>
>
> Thanks,
> M.D.
>
>
> On 8/1/2013 11:52 PM, Jeremy Rowley wrote:
>
> Thanks Moudrick.
>
>
>
> None of those documents require a certificate to include the anyEKU or no
> EKU.  Is there any software that actually requires the anyEKU or no EKU or
> is this just something that has happened over time with CAs?  If there isn’t
> an actual reason to put this in Qualified certs (other than CAs have
> included it in the past) then there isn’t a conflict, and we can move
> forward with the ballot.  Issuers of qualified certs will just need to start
> inserting the correct EKUs.  The problem (and related danger) is real enough
> that certs with anyEKU or no EKU should be covered by the BRs.
>
>
>
> Jeremy
>
>
>
> From: Moudrick M. Dadashov [mailto:md at ssc.lt]
> Sent: Thursday, August 01, 2013 11:39 AM
> To: jeremy.rowley at digicert.com
> Cc: 'Ryan Hurst'; 'CABFPub'
> Subject: Re: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> Jeremy,
>
> 1. ETSI TS 101 862 V1.3.3 (2006-01) Qualified Certificate profile
> 2. ETSI EN 319 412-5 V1.1.1 (2013-01) Electronic Signatures and
> Infrastructures (ESI); Profiles for Trust Service Providers issuing
> certificates; Part 5: Extension for Qualified Certificate profile
>
> also have a look:
>
> ETSI TS 119 412-2 V1.1.1 (2012-04) Electronic Signatures and Infrastructures
> (ESI); Profiles for Trust Service Providers issuing certificates; Part 2:
> Certificate Profile for certificates issued to natural persons.
>
> Search and download here:
>
> http://pda.etsi.org/pda/queryform.asp
>
> need to register, its free.
>
> Thanks.
> M.D.
>
> On 8/1/2013 7:18 PM, Jeremy Rowley wrote:
>
> Do you have a link for the profile?   None of the qualified cert profile
> recommendations or requirements I am aware of require the anyEKU or omission
> of the EKU.  They all say EKUS MUST be set in accordance with 5280.
>
>
>
> Jeremy
>
>
>
> From: public-bounces at cabforum.org [mailto:public-bounces at cabforum.org] On
> Behalf Of Moudrick M. Dadashov
> Sent: Thursday, August 01, 2013 9:25 AM
> To: Ryan Hurst
> Cc: 'CABFPub'
> Subject: Re: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> I see potential problem for ETSI Qualified SSL certificates, if key usage
> requirements remain mandatory as it is now with the Qualified el. signature
> certs.
>
> Thanks,
> M.D.
>
> On 8/1/2013 5:28 PM, Ryan Hurst wrote:
>
> There is nothing in the RFC that requires applications to treat the presence
> of the EKU as mandatory:
>
>
>
>    Certificate using
>
>    applications MAY require that the extended key usage extension be
>
>    present and that a particular purpose be indicated in order for the
>
>    certificate to be acceptable to that application.
>
>
>
> In fact the processing semantics defined in the RFC result in the behavior
> you see in all browsers today. That is if extended key usage is present the
> certificate is good for all usages, that if any key usage is present the
> certificate is only good for the specified usages.
>
>
>
>    If the extension is present, then the certificate MUST only be used
>
>    for one of the purposes indicated.  If multiple purposes are
>
>    indicated the application need not recognize all purposes indicated,
>
>    as long as the intended purpose is present.
>
>
>
> While this behavior does require CAs to be mindful of what applications
> require to understand the implications of their certificate profiles it is
> both standards compliant and the way things have been since the introduction
> of this extension.
>
>
>
> As for key usage; most libraries (certainly CryptoAPI) don’t pay attention
> to the key usage extension; that is with the exception of keyCertSign.
> Additionally its perfectly legitimate for an application to ignore the Key
> Usage field:
>
>
>
>    This extension indicates one or more purposes for which the certified
>
>    public key may be used, in addition to or in place of the basic
>
>    purposes indicated in the key usage extension.
>
>
>
> And the specification of the EKU of server authentication includes a set of
> specific KUs that are consistent with the extension:
>
>    id-kp-serverAuth             OBJECT IDENTIFIER ::= { id-kp 1 }
>
>    -- TLS WWW server authentication
>
>    -- Key usage bits that may be consistent: digitalSignature,
>
>    -- keyEncipherment or keyAgreement
>
>
>
> Given these facts I do not think it makes sense to include KU as part of the
> definition of scope.
>
>
>
> Ryan
>
> From: public-bounces at cabforum.org [mailto:public-bounces at cabforum.org] On
> Behalf Of Rijt, R.A. van de (Robert) - Logius
> Sent: Thursday, August 01, 2013 5:09 PM
> To: jeremy.rowley at digicert.com; 'Steve Roylance'
> Cc: 'CABFPub'
> Subject: Re: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> Ideally, only certificates that explicitly contain an EKU with serverauth
> would be considered SSL certs. All other certs should be dismissed. That
> would be in line with the RFC, but I realize this proposal might even be
> more impractical.
>
>
>
> I would suggest though to add to the current definition that only
> certificates that contain a KeyUsage with the digitalsignature and
> keyEncipherment and / or keyAgreement bits set, would be considered SSL
> certificates.  It just does not make sense to mandate that a personal
> certificate on a SSCD with KeyUsage non-repudation and no EKU would be
> considered an SSL certificate. That is not how I have interpreted RFC 5280.
>
>
>
> If the proposed definition is accepted all certificates with noEKU or anyEKU
> bits set will be governed by the BR. That means that all client certificates
> with those bits set, SSL or otherwise, will be governed by the rules of the
> BR. This in turn means that a client certificate must contain a FQDN, which
> it obviously cannot. In my view, adopting the proposed definition would lead
> to a situation where no client certificates can be issued under the roots
> present in the root programs, unless the burden of change is placed on those
> CAs issuing client certificates, forcing them to add keyusage bits to their
> certificates that are not compulsory through the RFCs. Furthermore, in many
> cases the anyEKU is relied on by software using client certificates.
>
>
>
> Regards,
>
> Robert
>
>
>
>
>
> Van: Jeremy Rowley [mailto:jeremy.rowley at digicert.com]
> Verzonden: donderdag 1 augustus 2013 14:06
> Aan: 'Steve Roylance'; Rijt, R.A. van de (Robert) - Logius
> CC: 'CABFPub'
> Onderwerp: RE: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> That is likely the way forward.  Mozilla can enable roots for “Web, code, or
> client” .  I assume the other browsers probably have a similar designation.
> If the root is disabled for “web” then the cert could not perform SSL and
> would not be considered enabled in the browser’s trust store (for SSL/TLS).
> The tweak to the proposed language would be nominal.
>
>
>
> From: public-bounces at cabforum.org [mailto:public-bounces at cabforum.org] On
> Behalf Of Steve Roylance
> Sent: Thursday, August 01, 2013 5:25 AM
> To: Rijt, R.A. van de (Robert) - Logius
> Cc: CABFPub
> Subject: Re: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> Hi Robert.
>
>
>
> Root program's have the ability to mark specific roots for specific uses
> therefore  you can still offer public trust but for a specific need. Maybe
> that's a way forward? As with Name Constraints it makes roots (or
> Subordinate CAs) less attractive as targets as their value to an attacker is
> decreased.
>
>
>
> Regards Steve
>
>
> Sent from my iPhone
>
>
> On 1 Aug 2013, at 12:04, "Rijt, R.A. van de (Robert) - Logius"
> <robert.vande.rijt at logius.nl> wrote:
>
> For qualified certificates under ETSI that need to be publicly trusted, a
> private root would not be an option. Moreover, developing a private,
> not-publicly trusted root and rolling out end-entity certificates takes
> time. I am talking about a year at least.
>
>
>
> I wonder if everyone else is realizing the impact on “non-SSL” certificates.
> Especially the CA’s not participating in the CABforum because they do not
> issue SSL certs (or thought they did not), but do have a publicly trusted
> root.
>
>
>
> Robert
>
>
>
> Van: Jeremy Rowley [mailto:jeremy.rowley at digicert.com]
> Verzonden: donderdag 1 augustus 2013 12:56
> Aan: Rijt, R.A. van de (Robert) - Logius; 'Ryan Hurst'
> CC: 'CABFPub'
> Onderwerp: RE: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> Which is why you will now have to issue these off a root not trusted by a
> participating browser. The safetynet is the problem since makes them an SSL
> cert.  I don’t think you can both have a safetynet like this and issue the
> cert from a trusted root.
>
>
>
> Jeremy
>
>
>
> From: Rijt, R.A. van de (Robert) - Logius
> [mailto:robert.vande.rijt at logius.nl]
> Sent: Thursday, August 01, 2013 4:53 AM
> To: Ryan Hurst; jeremy.rowley at digicert.com
> Cc: CABFPub
> Subject: RE: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
>
>
> Thanks for your reply, Jeremy. I conclude that with this new definition:
>
>
>
> 1. We are forcing everyone with public certificates to use the EKU;
>
> 2. We are forcing everyone with public certificates not to use the
> anyExtendedKeyUsage unless it is a SSL certificate;
>
> 3. thereby forcing everyone to spell out all the applicable EKUs one–by-one.
> My experience is that a lot of software cannot handle this, so the
> certificate cannot be used for the function intended. That is why the
> anyExtendedKeyUsage is often used as a safetynet.
>
>
>
> Robert
>
>
>
>
>
> Van: Ryan Hurst [mailto:ryan.hurst at globalsign.com]
> Verzonden: donderdag 1 augustus 2013 12:45
> Aan: jeremy.rowley at digicert.com
> CC: Rijt, R.A. van de (Robert) - Logius; CABFPub
> Onderwerp: Re: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> I concur with Jeremy's analysis.
>
> Ryan Hurst
>
> Chief Technology Officer
>
> GMO Globalsign
>
>
>
> twitter: @rmhrisk
>
> email: ryan.hurst at globalsign.com
>
> phone: 206-650-7926
>
>
>
> Sent from my phone, please forgive the brevity.
>
>
> On Aug 1, 2013, at 1:12 PM, "Jeremy Rowley" <jeremy.rowley at digicert.com>
> wrote:
>
> HI Rijt,
>
>
>
> I think the certificates you mentioned will (and should) qualify under the
> BRs if are issued from a root that is included in one of the adopting
> browser’s trust stores.
>
>
>
> Here’s my logic:
>
> 1)      Certs that don’t have an EKU or that include the anyEKU can be used
> as SSL certs, regardless of their intended purposes and should (arguably) be
> under the BRs.
>
> 2)      However, as you mentioned, many certificates assert the anyEKU, have
> noEKU, or even contain the server authentication EKU that are never intended
> to be used on a server.
>
> 3)      I believe the certificates referenced typically lack an FQDN.
> Including these certificates under the BR umbrella is problematic because
> the certificates can’t function in the intended manner and comply with the
> BRs.  The CN is an identifier for the equipment, which violates Section
> 9.2.2 of the BRs.
>
> 4)      Requiring an FQDN for inclusion in the BRs is not a way forward
> since that would make the sections on internal server names out of scope of
> the BRs. In fact, the identifier in these certificates is indistinguishable
> from and qualifies as an internal name, meaning the certificate presents all
> of the concerns previously expressed by PayPal.   Continuing to trust these
> certificates would be the same as not deprecating internal server name
> certificates
>
> 5)      Therefore, these certificates must either be included in the BRs,
> and include an FQDN, or need to be issued off of a non-publicly trusted root
> certificate.
>
>
>
> The position you expressed is why I wanted to raise the issue and why I
> think we need to resolve what the BRs actually cover.
>
>
>
> Jeremy
>
>
>
> From: public-bounces at cabforum.org [mailto:public-bounces at cabforum.org] On
> Behalf Of Rijt, R.A. van de (Robert) - Logius
> Sent: Thursday, August 01, 2013 3:05 AM
> To: 'CABFPub'
> Subject: Re: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> Although I understand the need for tightening the definition and I can
> follow the reasoning below to a certain point I feel that, instead of
> tightening it, the new definition seems to have broadened the scope. The
> vast majority of certificates issued under the Logius PKIoverheid root are
> not intended for the identification of SSL servers. However, roughly 90% of
> these certificates will now fall under this new definition. In the present
> version, the scope made clear that the BR only addressed certificates meant
> for servers.
>
>
>
> What about personal certificates on a SSCD that have no EKU or have an
> anyExtendedKeyUsage as a safetynet? Would these certificates suddenly by
> seen as SSL certificates although they are obviously not intended for
> servers? What about certificates issued to autonomous devices such as
> onboard computers in taxicabs or domestic gas meters, to name but two? Would
> these be considered SSL certificates if they have no EKU or the clientauth
> EKU in combination with anyExtendedKeyUsage?
>
>
>
> Regards,
>
> Robert
>
>
>
>
>
>
>
>
>
> Van: public-bounces at cabforum.org [mailto:public-bounces at cabforum.org] Namens
> Ryan Sleevi
> Verzonden: maandag 29 juli 2013 22:57
> Aan: Kelvin Yiu
> CC: CABFPub
> Onderwerp: Re: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
>
>
> They're still respected (for better or worse) by Apple, NSS, and Android.
>
> Even if that changed tomorrow, the fact that a significant portion of the
> deployed user base for those products may not upgrade immediately suggests
> it would be wise to keep them in scope - especially given that even few
> products implement Microsoft's EKU chaining behaviour for intermediates.
>
> On Jul 29, 2013 1:52 PM, "Kelvin Yiu" <kelviny at exchange.microsoft.com>
> wrote:
>
> I prefer to drop any mention of the MS or Netscape SGC OIDs. These OIDs have
> been obsolete for over a decade and have ceased to have any meaning on MS
> platforms since Windows 2000.
>
> Kelvin
>
> -----Original Message-----
> From: public-bounces at cabforum.org [mailto:public-bounces at cabforum.org] On
> Behalf Of Ryan Sleevi
> Sent: Friday, July 26, 2013 12:35 PM
> To: jeremy rowley
> Cc: CABFPub
> Subject: Re: [cabfpub] Ballot 108: Clarifying the scope of the baseline
> requirements
>
> Jeremy,
>
> If I might suggest a slight modification to the wording, which still leaves
> things a little ambiguous. "All root and intermediate certificates included
> in a browser's trust store" - AIUI, none of the browsers are really
> accepting intermediates to the trust store at this point.
>
> This was a sticky point when working on Mozilla's wording on this policy to.
> Luckily, the terminology for "Root CA" and "Subordinate CA"
> may be sufficient here to help clarify.
>
> "All root certificates included in a browser's trust store, all subordinate
> CA certificates signed by one of these root certificates, and all end-entity
> certificates that either lack any Extended Key Usage extension or contain an
> Extended Key Usage extension that contains one of the following:
> - Server Authentication (1.3.6.1.5.5.7.3.1)
> - anyExtendedKeyUsage (2.5.29.37.0)
> - Netscape Server Gated Cryptography (2.16.840.1.113730.4.1)
> - Microsoft Server Gated Cryptography (1.3.6.1.4.1.311.10.3.3) are expressly
> covered by these requirements."
>
> Note that Appendix B, 3.F lists other values as SHOULD NOT. However, that
> presumably only applies when a certificate is 'in scope' of the BRs, hence
> the restatement of potential EKUs that are relevant.
>
>
>
> On Fri, Jul 26, 2013 at 12:22 PM, Jeremy Rowley <jeremy.rowley at digicert.com>
> wrote:
>> Hi everyone,
>>
>>
>>
>> As mentioned on the phone call last week, CAs have claimed exemption
>> from the BRs because the definition of a publicly-trusted SSL certificate
>> is not
>> clear.   I would like to clarify the scope of the BRs to avoid confusion
>> on
>> what particular certificate contents are necessary to require
>> compliance.  I am looking for on endorser (Stephen Davidson has already
>> endorsed).
>>
>>
>>
>> The third paragraph of Section 1 of the baseline requirements is:
>>
>> "This version of the Requirements only addresses Certificates intended
>> to be used for authenticating servers  accessible through the
>> Internet. Similar requirements for code signing, S/MIME,
>> time-stamping, VoIP, IM, Web services, etc. may be covered in future
>> versions."
>>
>>
>>
>> I'd like to replace the above text with the following:
>>
>> "This version of the Baseline Requirements addresses all root,
>> intermediate, and end entity certificates that can be used in
>> publicly-trusted SSL handshakes.  All root and intermediate
>> certificates included in a browser's trust store and all end entity
>> certificates containing an extended key usage extension of Server
>> Authentication (1.3.6.1.5.5.7.3.1) are expressly covered by these
>> requirements. Similar requirements for code signing, S/MIME,
>> time-stamping, VoIP, IM, Web services, etc. may be covered in future
>> versions."
>>
>>
>>
>> I look forward to your comments.
>>
>>
>>
>> Jeremy
>>
>>
>> _______________________________________________
>> Public mailing list
>> Public at cabforum.org
>> https://cabforum.org/mailman/listinfo/public
>>
> _______________________________________________
> Public mailing list
> Public at cabforum.org
> https://cabforum.org/mailman/listinfo/public
>
>
>
> ________________________________
>
>
> Dit bericht kan informatie bevatten die niet voor u is bestemd. Indien u
> niet de geadresseerde bent of dit bericht abusievelijk aan u is toegezonden,
> wordt u verzocht dat aan de afzender te melden en het bericht te
> verwijderen. De Staat aanvaardt geen aansprakelijkheid voor schade, van
> welke aard ook, die verband houdt met risico's verbonden aan het
> elektronisch verzenden van berichten.
> This message may contain information that is not intended for you. If you
> are not the addressee or if this message was sent to you by mistake, you are
> requested to inform the sender and delete the message. The State accepts no
> liability for damage of any kind resulting from the risks inherent in the
> electronic transmission of messages. .
>
> _______________________________________________
> Public mailing list
> Public at cabforum.org
> https://cabforum.org/mailman/listinfo/public
>
>
>
> ________________________________
>
>
> Dit bericht kan informatie bevatten die niet voor u is bestemd. Indien u
> niet de geadresseerde bent of dit bericht abusievelijk aan u is toegezonden,
> wordt u verzocht dat aan de afzender te melden en het bericht te
> verwijderen. De Staat aanvaardt geen aansprakelijkheid voor schade, van
> welke aard ook, die verband houdt met risico's verbonden aan het
> elektronisch verzenden van berichten.
> This message may contain information that is not intended for you. If you
> are not the addressee or if this message was sent to you by mistake, you are
> requested to inform the sender and delete the message. The State accepts no
> liability for damage of any kind resulting from the risks inherent in the
> electronic transmission of messages. .
>
>
>
> ________________________________
>
>
> Dit bericht kan informatie bevatten die niet voor u is bestemd. Indien u
> niet de geadresseerde bent of dit bericht abusievelijk aan u is toegezonden,
> wordt u verzocht dat aan de afzender te melden en het bericht te
> verwijderen. De Staat aanvaardt geen aansprakelijkheid voor schade, van
> welke aard ook, die verband houdt met risico's verbonden aan het
> elektronisch verzenden van berichten.
> This message may contain information that is not intended for you. If you
> are not the addressee or if this message was sent to you by mistake, you are
> requested to inform the sender and delete the message. The State accepts no
> liability for damage of any kind resulting from the risks inherent in the
> electronic transmission of messages. .
>
> _______________________________________________
> Public mailing list
> Public at cabforum.org
> https://cabforum.org/mailman/listinfo/public
>
>
>
> ________________________________
>
>
> Dit bericht kan informatie bevatten die niet voor u is bestemd. Indien u
> niet de geadresseerde bent of dit bericht abusievelijk aan u is toegezonden,
> wordt u verzocht dat aan de afzender te melden en het bericht te
> verwijderen. De Staat aanvaardt geen aansprakelijkheid voor schade, van
> welke aard ook, die verband houdt met risico's verbonden aan het
> elektronisch verzenden van berichten.
> This message may contain information that is not intended for you. If you
> are not the addressee or if this message was sent to you by mistake, you are
> requested to inform the sender and delete the message. The State accepts no
> liability for damage of any kind resulting from the risks inherent in the
> electronic transmission of messages. .
>
>
>
>
>
> _______________________________________________
>
> Public mailing list
>
> Public at cabforum.org
>
> https://cabforum.org/mailman/listinfo/public
>
>
>
>
>
>
>
> _______________________________________________
> Public mailing list
> Public at cabforum.org
> https://cabforum.org/mailman/listinfo/public
>



More information about the Public mailing list