Guang Dong Certificate Authority (GDCA) root inclusion request

[Kathleen Wilson]

This request from Guangdong Certificate Authority (GDCA) is to include the "GDCA TrustAUTH R5 ROOT" certificate, turn on the Websites trust bit, and enabled EV treatment.

GDCA is a nationally recognized CA that operates under China’s Electronic Signature Law. GDCA’s customers are business corporations registered in mainland China, government agencies of China, individuals or mainland China citizens, servers of business corporations which have been registered in mainland China, and software developers.

The request is documented in the following bug:
https://bugzilla.mozilla.org/show_bug.cgi?id=1128392

And in the pending certificates list:
https://wiki.mozilla.org/CA:PendingCAs

Summary of Information Gathered and Verified:
https://bugzilla.mozilla.org/attachment.cgi?id=8749437

Noteworthy points:

* Root Certificate Download URL:
https://bugzilla.mozilla.org/attachment.cgi?id=8748933
https://www.gdca.com.cn/cert/GDCA_TrustAUTH_R5_ROOT.der

* The primary documents are provided in Chinese.

CA Document Repository: https://www.gdca.com.cn/customer_service/knowledge_universe/cp_cps/
http://www.gdca.com.cn/cp/cp
http://www.gdca.com.cn/cps/cps
http://www.gdca.com.cn/cp/ev-cp
http://www.gdca.com.cn/cps/ev-cps

Translations into English:
CP: https://bugzilla.mozilla.org/attachment.cgi?id=8650346
CPS: https://bugzilla.mozilla.org/attachment.cgi?id=8688749

* CA Hierarchy: This root certificate has internally-operated subordinate CAs
- GDCA TrustAUTH R4 SSL CA (issues 2048-bit SSL certs)
- GDCA TrustAUTH R4 Generic CA (issues 2048-bit individual certs)
- GDCA TrustAUTH R4 CodeSigning CA (issues 2048-bit CodeSigning certs)
- GDCA TrustAUTH R4 Extended Validation SSL CA (issues 2048-bit EV SSL certs)
- GDCA TrustAUTH R4 Extended Validation Code Signing CA (issues 2048-bit EV CodeSigning certs)

* This request is to turn on the Websites trust bit.

CPS section 3.2.5: For domain verification, GDCA needs to check the written materials which can be used to prove the ownership of corresponding domain provided by applicant. Meanwhile, GDCA should ensure the ownership of domain from corresponding registrant or other authoritative third-party databases. During the verification, GDCA needs to perform the following procedures:
1. GDCA should confirm that the domain's owner is certificate applicant based on the information queried from corresponding domain registrant or authoritative third-party database and provided by applicant.
2. GDCA should confirm that the significant information (such as document information of applicant) in application materials are consistent with the reply of domain's owner by sending email or making phone call based on the contact information (such as email, registrar, administrator's email published at this domain's website, etc.) queried from corresponding domain registrant or authoritative third-party database.
If necessary, GDCA also need to take other review measures to confirm the ownership of the domain name. Applicant can't refuse to the request for providing appropriate assistance.


* EV Policy OID: 1.2.156.112559.1.1.6.1

* Test Website: https://ev-ssl-test-1.95105813.cn/

* CRL URLs:
http://www.gdca.com.cn/crl/GDCA_TrustAUTH_R5_ROOT.crl
http://www.gdca.com.cn/crl/GDCA_TrustAUTH_R4_SSL_CA.crl
http://www.gdca.com.cn/crl/GDCA_TrustAUTH_R4_Extended_Validation_SSL_CA.crl

* OCSP URL:
http://www.gdca.com.cn/TrustAUTH/ocsp

* Audit: Annual audits are performed by PricewaterhouseCoopers Zhong Tian LLP according to the WebTrust criteria.
WebTrust CA: https://cert.webtrust.org/SealFile?seal=2024&file=pdf
WebTrust BR: https://cert.webtrust.org/SealFile?seal=2025&file=pdf
WebTrust EV: https://cert.webtrust.org/SealFile?seal=2026&file=pdf

* Potentially Problematic Practices: None Noted
(http://wiki.mozilla.org/CA:Problematic_Practices)

This begins the discussion of the request from Guangdong Certificate Authority (GDCA) to include the "GDCA TrustAUTH R5 ROOT" certificate, turn on the Websites trust bit, and enabled EV treatment. At the conclusion of this discussion I will provide a summary of issues noted and action items. If there are outstanding issues, then an additional discussion may be needed as follow-up. If there are no outstanding issues, then I will recommend approval of this request in the bug.

Kathleen

[Kathleen Wilson]

All,

I will greatly appreciate it if you will review and comment on this request from Guangdong Certificate Authority (GDCA) to include the "GDCA TrustAUTH R5 ROOT" certificate, turn on the Websites trust bit, and enabled EV treatment.

Thanks,
Kathleen

[Peter Bowen]

On Wed, Aug 3, 2016 at 2:45 PM, Kathleen Wilson <kwi...@mozilla.com> wrote:
> This request from Guangdong Certificate Authority (GDCA) is to include the "GDCA TrustAUTH R5 ROOT" certificate, turn on the Websites trust bit, and enabled EV treatment.

>
> * CA Hierarchy: This root certificate has internally-operated subordinate CAs
> - GDCA TrustAUTH R4 SSL CA (issues 2048-bit SSL certs)
> - GDCA TrustAUTH R4 Generic CA (issues 2048-bit individual certs)
> - GDCA TrustAUTH R4 CodeSigning CA (issues 2048-bit CodeSigning certs)
> - GDCA TrustAUTH R4 Extended Validation SSL CA (issues 2048-bit EV SSL certs)
> - GDCA TrustAUTH R4 Extended Validation Code Signing CA (issues 2048-bit EV CodeSigning certs)
>

> * Audit: Annual audits are performed by PricewaterhouseCoopers Zhong Tian LLP according to the WebTrust criteria.
> WebTrust CA: https://cert.webtrust.org/SealFile?seal=2024&file=pdf
> WebTrust BR: https://cert.webtrust.org/SealFile?seal=2025&file=pdf
> WebTrust EV: https://cert.webtrust.org/SealFile?seal=2026&file=pdf

Kathleen and team,

I reviewed the annual audit reports linked in your email, including
the auditor's opinion and the management assertions.

Good:
- The reports and management assertion include an English language version
- The English versions are authoritative (no qualification the Chinese
language version holds in case of conflict)
- The reports clearly state they use the International Standard on
Assurance Engagements (ISAE) 3000 standard
- The report and assertion use the current version of the criteria
- The assertions include details of each CA in the appendix, including
an indication of whether it is a Root CA and a cryptographic hash
(fingerprint) associated with each CA

Opportunties for Improvement:
- The basic WebTrust Report and assertion do not specify location(s)
where services are provided. Other reports do indicate the services
are provided in/from China.
- The opinions do not specify the list of Certification Authorities in scope
- The reports and asssertions do not specify the versions of the CP and CPS
- The management assertion appendixes mix keys and certificates. The
certificate is not the interesting part; the CA Distinguished Name
(DN), type (Root or not), Key, and Key ID are the interesting parts.
- The BR report repeats a bullet under "Maintained effective controls
to provide reasonable assurance that:"

Bad:
- The basic WebTrust for CA Report does not cover controls that
provide assurance that subordinate CA certificate requests are
accurate, authenticated, and approved (the management assertion does,
so this indicates the auditor might have found issues with the
controls)
- The basic WebTrust for CA Management assertion does not include
Subordinate CA [cross-]certification in the list of CA services
- The basic WebTrust for CA Management assertion does not include
"Subordinate CA Certificate Lifecycle Management Controls" in the list
of portions of criteria used
- After the reporting period ended, GDCA issued at least two new
subordinate CA certificates from the R5 root. These use the
organization name of "Global Digital Cybersecurity Authority Co.,
Ltd." and have keys and key IDs that are identical to those found in
CA certificates for GUANG DONG CERTIFICATE AUTHORITY CO.,LTD. This is
problematic as re-use of key IDs with different issuer names causes
problems on some platforms. Additionally the separate DN means it is
out of scope for the submitted report. Combined with the lack of
audited controls around subordinate CA management, CAs outside of the
scope of the report may be a significant concern.
- The Baseline + Network Security Requirements report and management
assertion only covers two of the CAs. However the cross-certs issued
by the root to the subordinate CAs do not include EKU constraints, so
the subordinate CAs are capable of issuing server authentication
("SSL") certificates. The assertion and report should include all CAs
that are capable of issuing ser authentication certificates.
- The Baseline report does not provide an option that GDCA "maintained
effective controls to provide reasonable assurance that it meets the
Network and Certificate System Security Requirements as set forth by
the CA/Browser Forum"
- The Extended Validation report and management assertion attempts to
merge the Extended Validation SSL and Extended Validation Code Signing
criteria. These should be distinct reports. As writtten, the report
fails to adequately cover the EVCS critera.

The WebTrust/PKI Task Force has published a draft set of illustrative
reports to use as a basis
(http://www.webtrust.org/practitioner-qualifications/item83253.pdf),
so it should be faily easy to resolve the missing bits.

Thanks,
Peter

[wangs...@gmail.com]

Dear Peter, Thanks for your comments! We think that there are some good suggestions for our work. We’ll take notes and do better in our future work.

We have discussed these questions with our auditor. Here are our reply to your comments:

Opportunties for Improvement:
- The basic WebTrust Report and assertion do not specify location(s) where services are provided. Other reports do indicate the services are provided in/from China.

Reply: It’s not a mandatory requirement but a good suggestion, we’ve told our auditor and they would use the new report template published by WebTrust/PKI Task Force in next year's audit which will specify the location.

- The opinions do not specify the list of Certification Authorities in scope

Reply: Our company has changed the name from “GUANG DONG CERTIFICATE AUTHORITY CO.,LTD.” to “Global Digital Cybersecurity Authority Co., Ltd.” in this year. We do not have any external sub-CAs, so it’s no need to list the CAs in scope.
https://www.gdca.com.cn/about_gdca/CTrends/new/Notification-of-Company-Name-Change/
https://www.gdca.com.cn/about_gdca/CTrends/new/-00119/

- The reports and asssertions do not specify the versions of the CP and CPS

Reply: It’s not a mandatory requirement but a good suggestion, we’ve told our auditor and they would use the new report template published by WebTrust/PKI Task Force in next year's audit which will specify the version of the CP and CPS.

- The management assertion appendixes mix keys and certificates. The certificate is not the interesting part; the CA Distinguished Name (DN), type (Root or not), Key, and Key ID are the interesting parts.

Reply: It’s a good suggestion. We may consider to specify them in our future assertion.

- The BR report repeats a bullet under "Maintained effective controls to provide reasonable assurance that:"

Reply: We’ve confirmed with the auditor and it’s just a clerical error in auditor’s report.

Bad:
- The basic WebTrust for CA Report does not cover controls that provide assurance that subordinate CA certificate requests are accurate, authenticated, and approved (the management assertion does, so this indicates the auditor might have found issues with the controls)

Reply: We've communicated with our auditor, they said that in the period of time report, we did not generate any subordinate certificates, which means that no subordinate certificate request happened during the audit period. So that the subordinate certificate request did not be mentioned in the audit report.

- The basic WebTrust for CA Management assertion does not include Subordinate CA [cross-]certification in the list of CA services.
Reply: We do not have any external sub-CAs, so it’s no need to include the cross-certs in the list.

- The basic WebTrust for CA Management assertion does not include "Subordinate CA Certificate Lifecycle Management Controls" in the list of portions of criteria used

Reply: We do not have any external sub-CAs, so it’s not applicable.

- After the reporting period ended, GDCA issued at least two new subordinate CA certificates from the R5 root. These use the organization name of "Global Digital Cybersecurity Authority Co., Ltd." and have keys and key IDs that are identical to those found in CA certificates for GUANG DONG CERTIFICATE AUTHORITY CO.,LTD. This is problematic as re-use of key IDs with different issuer names causes problems on some platforms. Additionally the separate DN means it is out of scope for the submitted report. Combined with the lack of audited controls around subordinate CA management, CAs outside of the scope of the report may be a significant concern.

Reply: Our company has changed the name from “GUANG DONG CERTIFICATE AUTHORITY CO.,LTD.” to “Global Digital Cybersecurity Authority Co., Ltd.” in this year, which was after the period in time audit. We have informed the Mozilla of the name change in advance. We also announced the name change to the public in our official website.

- The Baseline + Network Security Requirements report and management assertion only covers two of the CAs. However the cross-certs issued by the root to the subordinate CAs do not include EKU constraints, so the subordinate CAs are capable of issuing server authentication ("SSL") certificates. The assertion and report should include all CAs that are capable of issuing ser authentication certificates.

Reply: We do not have any external sub-CAs.

- The Baseline report does not provide an option that GDCA "maintained effective controls to provide reasonable assurance that it meets the Network and Certificate System Security Requirements as set forth by the CA/Browser Forum"

Reply: The Baseline report issued by our auditor follows the “Webtrust Principles and Criteria for Certification Authorities – SSL Baseline with Network Security - Version 2.0” which is based on “Baseline Requirements for the Issuance and Management of Publicly – Trusted Certificates - Version 1.1.6” and “Network and Certificate Systems Security Requirement - Version 1.0”.(Please refer to http://www.webtrust.org/homepage-documents/item79806.pdf) So the SSL report has covered the assurance of the Network and Certificate System Security Requirement. We’ve also communicated with the auditor and confirmed that their audit work include the Network and Certificate System Security Requirement.

- The Extended Validation report and management assertion attempts to merge the Extended Validation SSL and Extended Validation Code Signing criteria. These should be distinct reports. As writtten, the report fails to adequately cover the EVCS critera.

Reply: It’s not a mandatory requirement before the publish of new report template. We’ve communicated with auditor and confirmed they have covered the audit work for EV SSL and EVCS in the audit process, and they would separate the EV SSL and EV Code Signing in their future report.

[Peter Bowen]

On Mon, Sep 19, 2016 at 1:56 AM, <wangs...@gmail.com> wrote:
> Dear Peter, Thanks for your comments! We think that there are some good suggestions for our work. We’ll take notes and do better in our future work.
>
> We have discussed these questions with our auditor. Here are our reply to your comments:
>

> - The basic WebTrust for CA Report does not cover controls that provide assurance that subordinate CA certificate requests are accurate, authenticated, and approved (the management assertion does, so this indicates the auditor might have found issues with the controls)

> Reply: We've communicated with our auditor, they said that in the period of time report, we did not generate any subordinate certificates, which means that no subordinate certificate request happened during the audit period. So that the subordinate certificate request did not be mentioned in the audit report.
>
> - The basic WebTrust for CA Management assertion does not include Subordinate CA [cross-]certification in the list of CA services.
> Reply: We do not have any external sub-CAs, so it’s no need to include the cross-certs in the list.
>

> - The basic WebTrust for CA Management assertion does not include "Subordinate CA Certificate Lifecycle Management Controls" in the list of portions of criteria used

> Reply: We do not have any external sub-CAs, so it’s not applicable.

You have one or more CAs that have a non-zero path constraint.
Therefore they are able to issue certificates to subordinate CAs,
whether operated by your organization or externally operated. As such,
it is important that the CA have controls around issuing CA
certificates and that the auditor has reasonable assurance the
controls function as designed.

> - After the reporting period ended, GDCA issued at least two new subordinate CA certificates from the R5 root. These use the organization name of "Global Digital Cybersecurity Authority Co., Ltd." and have keys and key IDs that are identical to those found in CA certificates for GUANG DONG CERTIFICATE AUTHORITY CO.,LTD. This is problematic as re-use of key IDs with different issuer names causes problems on some platforms. Additionally the separate DN means it is out of scope for the submitted report. Combined with the lack of audited controls around subordinate CA management, CAs outside of the scope of the report may be a significant concern.

> Reply: Our company has changed the name from “GUANG DONG CERTIFICATE AUTHORITY CO.,LTD.” to “Global Digital Cybersecurity Authority Co., Ltd.” in this year, which was after the period in time audit. We have informed the Mozilla of the name change in advance. We also announced the name change to the public in our official website.

I understand it was a name change. However you issued new CA
certificates with the new name which means you do issue CA
certificates from time to time.

> - The Baseline + Network Security Requirements report and management assertion only covers two of the CAs. However the cross-certs issued by the root to the subordinate CAs do not include EKU constraints, so the subordinate CAs are capable of issuing server authentication ("SSL") certificates. The assertion and report should include all CAs that are capable of issuing server authentication certificates.

> Reply: We do not have any external sub-CAs.

The question is not whether you have external sub-CAs, but whether all
the CAs that are subordinate to your root have controls around
issuance of server authentication certificates.

> - The Baseline report does not provide an option that GDCA "maintained effective controls to provide reasonable assurance that it meets the Network and Certificate System Security Requirements as set forth by the CA/Browser Forum"

> Reply: The Baseline report issued by our auditor follows the “Webtrust Principles and Criteria for Certification Authorities – SSL Baseline with Network Security - Version 2.0” which is based on “Baseline Requirements for the Issuance and Management of Publicly – Trusted Certificates - Version 1.1.6” and “Network and Certificate Systems Security Requirement - Version 1.0”.(Please refer to http://www.webtrust.org/homepage-documents/item79806.pdf) So the SSL report has covered the assurance of the Network and Certificate System Security Requirement. We’ve also communicated with the auditor and confirmed that their audit work include the Network and Certificate System Security Requirement.

I'm glad that the auditor has confirmed to you they did include the
Network Security criteria. However without it being included in their
report, it is not possible for a third party to have assurance they
did.

> - The Extended Validation report and management assertion attempts to merge the Extended Validation SSL and Extended Validation Code Signing criteria. These should be distinct reports. As writtten, the report fails to adequately cover the EVCS critera.

> Reply: It’s not a mandatory requirement before the publish of new report template. We’ve communicated with auditor and confirmed they have covered the audit work for EV SSL and EVCS in the audit process, and they would separate the EV SSL and EV Code Signing in their future report.

The EVCS audit does not build on the SSL BR audit, so it is important
that the report calls out that the CA maintained effective controls to
provide reasonable assurance that: requests for EV CS Signing
Authority and EV CS Timestamp Authority certificates are properly
authenticated; and certificates issued to EV CS Signing Authorities
and EV CS Timestamp Authorities are not valid for a period longer than
specified by the CA/Browser Forum. The report here does not call out
that the auditor confirmed there are controls to ensure these are
true.

However, Mozilla no longer grants code signing trust to CAs, so this
is outside of the scope of this inclusion request.

Thanks,
Peter

[wangs...@gmail.com]

在 2016年9月17日星期六 UTC+8上午5:38:29，Percy写道：

> https://www.ssllabs.com/ssltest/analyze.html?d=www.gdca.com.cn
> This server is vulnerable to the OpenSSL Padding Oracle vulnerability (CVE-2016-2107) and insecure. Grade set to F.
>
> Maybe someone who has more expertise than me could take a look at this?
Thank you for your review of our server. We had fixed this issue. Now the server returns "If trust issues are ignored: A". Please make a validation.

[wangs...@gmail.com]

> Dear Peter, Thanks for your comments! We think that there are some good suggestions for our work. We’ll take notes and do better in our future work. >> We have discussed these questions with our auditor. Here are our reply to your comments: >
> - The basic WebTrust for CA Report does not cover controls that provide assurance that subordinate CA certificate requests are accurate, authenticated, and approved (the management assertion does, so this indicates the auditor might have found issues with the controls) > Reply: We've communicated with our auditor, they said that in the period of time report, we did not generate any subordinate certificates, which means that no subordinate certificate request happened during the audit period. So that the subordinate certificate request did not be mentioned in the audit report. >> - The basic WebTrust for CA Management assertion does not include Subordinate CA [cross-]certification in the list of CA services. > Reply: We do not have any external sub-CAs, so it’s no need to include the cross-certs in the list. >> - The basic WebTrust for CA Management assertion does not include "Subordinate CA Certificate Lifecycle Management Controls" in the list of portions of criteria used > Reply: We do not have any external sub-CAs, so it’s not applicable.
You have one or more CAs that have a non-zero path constraint. Therefore they are able to issue certificates to subordinate CAs, whether operated by your organization or externally operated. As such, it is important that the CA have controls around issuing CA certificates and that the auditor has reasonable assurance the controls function as designed.

Reply: We’ve communicated with our auditor and knew that they always performed design effectiveness audit to all of our controls, include controls around issuing CA certificates. As there is no subordinate certificate request happened during the audit period, it was not mentioned in the audit report. It’s a good suggestion, they would specify the controls around issuing CA certificates in the future report no matter if it happens.

> - After the reporting period ended, GDCA issued at least two new subordinate CA certificates from the R5 root. These use the organization name of "Global Digital Cybersecurity Authority Co., Ltd." and have keys and key IDs that are identical to those found in CA certificates for GUANG DONG CERTIFICATE AUTHORITY CO.,LTD. This is problematic as re-use of key IDs with different issuer names causes problems on some platforms. Additionally the separate DN means it is out of scope for the submitted report. Combined with the lack of audited controls around subordinate CA management, CAs outside of the scope of the report may be a significant concern. > Reply: Our company has changed the name from “GUANG DONG CERTIFICATE AUTHORITY CO.,LTD.” to “Global Digital Cybersecurity Authority Co., Ltd.” in this year, which was after the period in time audit. We have informed the Mozilla of the name change in advance. We also announced the name change to the public in our official website.
I understand it was a name change. However you issued new CA certificates with the new name which means you do issue CA certificates from time to time.

Reply: The issuing of new CA certificates was after the period of audit report. We’ve informed Mozilla of the situation that we’ve issued new CA certificates which only changed the DN because of our company’s name change to facilitate our customers' use. During the issuing process, we also consulted our auditor about the format of new CA certificates. Also, if we don't change the DN of certificates, it will be difficult for browser users to confirm the identity of us when they visit the website whose certificate was signed by us because the CA name in certificate is “Guangdong certificate…” while our actual name is “Global Digital…”. It may cause unnecessary puzzle to browser users.

> - The Baseline + Network Security Requirements report and management assertion only covers two of the CAs. However the cross-certs issued by the root to the subordinate CAs do not include EKU constraints, so the subordinate CAs are capable of issuing server authentication ("SSL") certificates. The assertion and report should include all CAs that are capable of issuing server authentication certificates.
> Reply: We do not have any external sub-CAs.

The question is not whether you have external sub-CAs, but whether all the CAs that are subordinate to your root have controls around issuance of server authentication certificates.

Reply: Our CA system has realized the control that only appointed CA certificates can issue SSL certificates. We’ve communicated with our auditor and confirmed that the control effectiveness has been covered in their audit work. So only CA certificates mentioned in the management assertion can issuing SSL certificates.

> - The Baseline report does not provide an option that GDCA "maintained effective controls to provide reasonable assurance that it meets the Network and Certificate System Security Requirements as set forth by the CA/Browser Forum" > Reply: The Baseline report issued by our auditor follows the “Webtrust Principles and Criteria for Certification Authorities – SSL Baseline with Network Security - Version 2.0” which is based on “Baseline Requirements for the Issuance and Management of Publicly – Trusted Certificates - Version 1.1.6” and “Network and Certificate Systems Security Requirement - Version 1.0”.(Please refer to http://www.webtrust.org/homepage-documents/item79806.pdf) So the SSL report has covered the assurance of the Network and Certificate System Security Requirement. We’ve also communicated with the auditor and confirmed that their audit work include the Network and Certificate System Security Requirement.
I'm glad that the auditor has confirmed to you they did include the Network Security criteria. However without it being included in their report, it is not possible for a third party to have assurance they did.

Reply: The audit criteria “Webtrust Principles and Criteria for Certification Authorities – SSL Baseline with Network Security - Version 2.0” has included the “Network and Certificate Systems Security Requirement” which is clearly listed in the criteria. So the baseline report has covered the audit to “Network and Certificate Systems Security Requirement” and the third party can easily found and confirm it.

> - The Extended Validation report and management assertion attempts to merge the Extended Validation SSL and Extended Validation Code Signing criteria. These should be distinct reports. As writtten, the report fails to adequately cover the EVCS critera. > Reply: It’s not a mandatory requirement before the publish of new report template. We’ve communicated with auditor and confirmed they have covered the audit work for EV SSL and EVCS in the audit process, and they would separate the EV SSL and EV Code Signing in their future report.
The EVCS audit does not build on the SSL BR audit, so it is important that the report calls out that the CA maintained effective controls to provide reasonable assurance that: requests for EV CS Signing Authority and EV CS Timestamp Authority certificates are properly authenticated; and certificates issued to EV CS Signing Authorities and EV CS Timestamp Authorities are not valid for a period longer than specified by the CA/Browser Forum. The report here does not call out that the auditor confirmed there are controls to ensure these are true. However, Mozilla no longer grants code signing trust to CAs, so this is outside of the scope of this inclusion request.

Reply: understood. We’ve communicated with our auditor and confirmed that they would separate the EV SSL and EV Code Signing in their future report.

[Andrew R. Whalley]

Hello,

I have completed a read through of the English translations of the CP
(v1.2) and CPS (v4.1). Before I post my comments I wanted to see if there
were any more recent translations? It looks like the local language
versions are 1.4 and 4.3 respectively.

Many thanks,

Andrew

> _______________________________________________
> dev-security-policy mailing list
> dev-secur...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-security-policy
>

[wangs...@gmail.com]

在 2016年9月27日星期二 UTC+8上午4:15:00，Andrew R. Whalley写道：

Yes, we have new version translations. We have uploaded to Bug 1128392.
CP V1.4: https://bug1128392.bmoattachments.org/attachment.cgi?id=8795090
CPS V4.3: https://bug1128392.bmoattachments.org/attachment.cgi?id=8795091
EV CP V1.2: https://bug1128392.bmoattachments.org/attachment.cgi?id=8795093
EV CPS V1.3: https://bug1128392.bmoattachments.org/attachment.cgi?id=8795094

[Andrew R. Whalley]

Hello,

Thank you for the links. I note, however, that there's at least one
difference between the native language version and the English translation:

http://www.gdca.com.cn/cps/cps version 4.3 has a section 4.2.4 covering
CAA.
https://bug1128392.bmoattachments.org/attachment.cgi?id=8795091 version 4.3
in English has no such section.

The fact there's a discrepancy is rather worrying. Could you please check
and let me know if there are any other substantive differences between the
Chinese and English versions?

Cheers,

Andrew

On Mon, Sep 26, 2016 at 7:17 PM, <wangs...@gmail.com> wrote:

> 在 2016年9月27日星期二 UTC+8上午4:15:00，Andrew R. Whalley写道：

> Yes, we have new version translations. We have uploaded to Bug 1128392.
> CP V1.4: https://bug1128392.bmoattachments.org/attachment.cgi?id=8795090
> CPS V4.3: https://bug1128392.bmoattachments.org/attachment.cgi?id=8795091
> EV CP V1.2: https://bug1128392.bmoattachments.org/attachment.
> cgi?id=8795093
> EV CPS V1.3: https://bug1128392.bmoattachments.org/attachment.
> cgi?id=8795094

[Han Yuwei]

在 2016年10月20日星期四 UTC+8上午5:27:42，Andrew R. Whalley写道：

My English is not good enough, maybe my translation can't represent the original Chinese meaning accurately.
Any question would be answered about the meaning.
All information are based on version 4.3
I will only note the translation of Chinese version which is different from English version.

1.1.1

Chinese version have a additional statement:

After rename of GDCA, The property, debt, rights, and business of "Guangdong Digital Certificate Authority Co. LTD" would be transfered to "Global Digital Cyversecurity Authority CO., LTD." Any contracts signed by "Guangdong Digital Certificate Authority Co. LTD" would also be transfered to "Global Digital Cyversecurity Authority CO., LTD."

1.1.2
Para.2 have a link http://www.gdca.com.cn/TrustAUTH/ returned 404 at 20OCT2016 13:50Z, so I can't verify the CP.
Chinese version missed the detail of Object identifier section

1.1.3
Chinese version: Currently, GDCA has 6 root certificates, including ROOTCA (RSA), GDCA ROOT CA, ROOTCA(SM2), GDCA TrustAUTH R5 ROOT, 数安时代 R5 ROOT (CN is Chinese), GDCA TrustAUTH E5 ROOT

1) About GDCA TrustAUTH R2 CA which will expire at 15DEC2018, from 15DEC2016, GDCA will no longer use it to issue subscriber certificates.
2) GDCA ROOT CA will expire on 11DEC2024.
4) GDCA TrustAUTH R5 would issue EV certficates
4) 5) 6) section is totally different from English version. I can't translate it all.

1.2
Chinese version missed the OID section.

1.3
The Chinese version's word is different. I think PKI should refer to Public Key Infrastructure. Am I right?

1.4.1
Chinses version have addtional statements.it said something about EV SSL Certificates.But I think it is not important.

1.4.1.1
Addtional type of individual certficates: Type III and Type IV which require more validations.

1.4.1.2
GDCA will NOT issue any Type I and II Organization certificate, only issue Type III and IV.

1.4.1.4
4 types: EV,OV,IV,DV. EV SSL would follow another CPS.

1.4.1.5
Some differences.Not important

3.1.1
For SSL cerificate......and a primary domain name or IP address shall be used as CN.

issuer's DN: O: Global Digital Cyversecurity Authority CO., LTD. or GDCA Certificate Authority

3.1.5

The first applicant of this DN shall govern, later applicant would be distinguished by addtional information.

3.2.2 Title: Authentication of Individual Identity which should be 3.2.3's title.
3.2.2 and 3.2.3 cannot be compared because Chinese version depends on different type of certificates.

3.2 is totally a mess.

4.2.4
Not availabe. GDCA don't do CAA validation.

4.7.1
not allowed to update key: Type I & II individual certificate,Equipment certificate,SSL certificate, Code Signing certficate.

6.3.3
- For RSA2048 SSL cert and Code Signing cert,ECC 256bit SSL and Code Signing cert, max period of keypair usage is 39 months

7.1.3
sha1RSA,sha256RSA and sha256ECDSA

7.2.2
Signing algorithm: sha1RSA sha256RSA sha256ECDSA SM2 ECC

Section 9 which is about law is too hard for me. I would only pick up something I can understand.

9.2.1
compensation will not exceed:
800CNY for Individual cert.
4,000CNY for Organzation cert.
8,000CNY for Equipment cert.
200,000CNY for Code Signing cert.
500,000CNY for SSL cert.


Appendix:
GDCA TrushAUTH R4 EV SSL CA & GDCA TrustAUTH R4 EV CodeSigning CA's information will be discloused in GDCA EV CPS
GDCA TrustAUTH R4 IV SSL CA (SHA1=78AEA851A31B0F049AF02CD0F2AD9140604FA7A3)
GDCA TrustAUTH R4 DV SSL CA (SHA1=30184A5B924E679E7A91329317D0560F587E697B)
GDCA TrustAUTH R4 Primer CA (SHA1=14C2B33BBF6EBD84FCA7015413EBD0433E171A98)
some Chinese CN CAs
some E5 CAs

[wangs...@gmail.com]

在 2016年10月21日星期五 UTC+8上午10:52:42，Percy写道：
> Thanks for bringing the discrepancy into our attention.
> Even the cover page of the English and Chinese version of CPS are dated differently.
>
> English
> Global Digital Cybersecurity Authority
> CO., LTD.
> Certification Practice Statement (CPS) Version: V4.3
> Effective Date: July 1, 2016
>
> Chinese
> 数安时代科技股份有限公司
> 电子认证业务规则
> 版本:V4.3
> 生效日期:2016 年 8 月 1 日 (Effective date: August 1st, 2016)
>
>
> In 1.1.3 3) the Chinese version shows ROOTCA(SM2) - Guangdong Certificate Authority(SM2) while the English version shows ROOTCA(SM2) - SM2 CA Certificate.
>
> I saw 4) included in the English version about 1.1.3 5) 数安时代 R5 根 CA and 6) GDCA TrustAUTH E5 ROOT
> are completely missing in the English version.
>
> I translated the 6) section below.
>
> GDCA TrustAUTH E5 ROOT uses ECC, with the root key length 384-bit . It has 7 sub-CA. 1)GDCA TrustAUTH E4 EV SSL CA with key length 256-bit and it signs 256-bit EV SSL servers. 2)GDCA TrustAUTH E4 OV SSL CA and it signs 256-bit OV SSL certs for servers. (3)GDCA TrustAUTH E4 IV SSL CA，256-bit key and signs 256-bit IV SSL certs for servers; (4)GDCA TrustAUTH E4 DV SSL CA，256-bit key and signs 256-bit DV SSL certs for servers; (5)GDCA TrustAUTH E4 CodeSigning CA 256-bit key，and signs 256-bit code certs;(6)GDCA TrustAUTH E4 Generic CA, 256-bit，signed 256-bit certs for organizations and individuals ;(7)GDCA TrustAUTH E4 Primer CA， 256-bit，and signs 256-bit personal certs.
>
> GDCA TrustAUTH E5 ROOT will expire on 2040 Dec 31st.
> GDCA TrustAUTH E4 EV SSL CA will expire on 2030 Dec 31st. From 2027 Jan,1st ，no new certs will be signed with it.
> …( More expiration date stuff)
>
> GDCA TrustAUTH R5 ROOT 、数安时代 R5 根 CA 证书、GDCA
> TrustAUTH E5 ROOT’s intermediate certs: GDCA conforms to the latest version of CA/Browser Forum Baseline Requirements for the Issuance and Management of Publicly Trusted SSL Digital Certificates published at www.cabforum.org. In the event that a discrepancy arises between interpretations of this document and Baseline Requirement, the Baseline Requirement shall govern.
>
> This is as far as I read. There are probably many more inconsistencies as Yuwei pointed out.
> I suggest Mozilla ask a staff who know Chinese to fully translate the Chinese CPS yourself and compare with the provided English CPS
Hi Andrew,Yuwei and Percy:
Thank you for your reviews of our CP and CPS.
The effective date of Chinese version was August 1 while the English version was July 1. We are now translating a new English version which match the Chinese version. We will upload a new English version next week when the translation is all complete.

[wangs...@gmail.com]

在 2016年10月21日星期五 UTC+8上午12:15:00，Han Yuwei写道：

Thanks again for Yuwei to list the major differences between the Chinese version and the English version.

>1.1.2
>Para.2 have a link http://www.gdca.com.cn/TrustAUTH/ returned 404 at 20OCT2016 13:50Z, so I can't verify the CP.
>Chinese version missed the detail of Object identifier section

The link is now http://www.gdca.com.cn/cp/cp

>1.1.3
>Chinese version: Currently, GDCA has 6 root certificates, including ROOTCA (RSA), GDCA ROOT CA, ROOTCA(SM2), GDCA TrustAUTH R5 ROOT, 数安时代 R5 ROOT (CN is Chinese), GDCA TrustAUTH E5 ROOT

>1) About GDCA TrustAUTH R2 CA which will expire at 15DEC2018, from 15DEC2016, GDCA will no longer use it to issue subscriber certificates.
>2) GDCA ROOT CA will expire on 11DEC2024.
>4) GDCA TrustAUTH R5 would issue EV certficates
>4) 5) 6) section is totally different from English version. I can't translate it all.

The 4) section is about the GDCA TrustAUTH R5 ROOT and sub-CAs
The 5) section is about the 数安时代 R5 ROOT and sub-CAs
The 6) section is about the GDCA TrustAUTH E5 ROOT and sub-CAs

>1.2
>Chinese version missed the OID section.

1.2. Document Name and Identification
In this document called "Global Digital Cybersecurity Authority CO., LTD. Certification Practice Statement" (abbreviated as “GDCA CPS”), CPS is equivalent to the document name and applicable name defined in this section.
The object identifier (OID) of certificates applied to the project of Hong Kong-Guangdong mutual recognition in this CPS are consistent with "Certificate Policy for Hong Kong-Guangdong mutual recognition of electronic signature certificates” while other are consistent with “GDCA Certificate Policy” (abbreviated as “GDCA CP”).

>3.2 is totally a mess.

The 3.2 section of the Chinese version is different from the English version now. Please see it in the new English version next week.

>Appendix:
>GDCA TrushAUTH R4 EV SSL CA & GDCA TrustAUTH R4 EV CodeSigning CA's information will be discloused in GDCA EV CPS
>GDCA TrustAUTH R4 IV SSL CA (SHA1=78AEA851A31B0F049AF02CD0F2AD9140604FA7A3)
>GDCA TrustAUTH R4 DV SSL CA (SHA1=30184A5B924E679E7A91329317D0560F587E697B)
>GDCA TrustAUTH R4 Primer CA (SHA1=14C2B33BBF6EBD84FCA7015413EBD0433E171A98)
>some Chinese CN CAs
>some E5 CAs

GDCA TrustAUTH R5 ROOT SHA1 digest = 0f 36 38 5b 81 1a 25 c3 9b 31 4e 83 ca e9 34 66 70 cc 74 b4
GDCA TrustAUTH R4 EV SSL CA See “GDCA EV CPS”
GDCA TrustAUTH R4 EV CodeSigning CA See “GDCA EV CPS”
GDCA TrustAUTH R4 OV SSL CA SHA1 digest = c3 4a d6 45 d5 79 1c 5f 22 e7 33 d7 53 47 08 15 85 75 6c 2d
GDCA TrustAUTH R4 IV SSL CA SHA1 digest = 78 ae a8 51 a3 1b 0f 04 9a f0 2c d0 f2 ad 91 40 60 4f a7 a3
GDCA TrustAUTH R4 DV SSL CA SHA1 digest = 30 18 4a 5b 92 4e 67 9e 7a 91 32 93 17 d0 56 0f 58 7e 69 7b
GDCA TrustAUTH R4 CodeSigning CA SHA1 digest = fc 6d cb 06 a5 5b ff 76 83 64 27 5b 29 d6 4f 7c 3a a9 cf b4
GDCA TrustAUTH R4 Generic CA SHA1 digest =6f ed 83 eb e1 83 cc 71 d0 ed e1 2a e8 77 e0 df 98 96 1f 24
GDCA TrustAUTH R4 Primer CA SHA1 digest =14 c2 b3 3b bf 6e bd 84 fc a7 01 54 13 eb d0 43 3e 17 1a 98

2 New Root:

数安时代R5根CA证书 SHA1 digest = 23 eb 1b a4 64 71 a1 e7 e9 f2 db 57 01 fe f8 f2 f8 0c aa e9
数安时代R4 EV 服务器 See “GDCA EV CPS”
数安时代R4 OV 服务器证书 CA SHA1 digest = 93 92 5b 05 17 30 05 86 fd 2c 45 eb 18 6e 00 9e b9 75 a5 d0
数安时代R4 IV 服务器证书 CA SHA1 digest = 10 b8 fb 9a d2 50 32 6a ee fb 05 ad da 9d 3a 2b bb bd 5d bf
数安时代R4 DV 服务器证书 CA SHA1 digest = 01 ad 04 cd e1 05 56 23 4a f6 6f a0 e6 64 f3 a6 18 80 4d f5
数安时代R4 代码签名证书 CA SHA1 digest = 4f be 54 bc 70 8e b1 2a 11 86 dd 79 aa ff e7 95 f8 ad c6 e9
数安时代R4 普通订户证书 CA SHA1 digest = 07 33 29 cb 53 b1 86 36 25 38 1b fb 48 a0 43 a7 b1 fe 28 6f
数安时代R4 基础订户证书 CA SHA1 digest = e5 da 52 2d 5f 38 7a 6e 72 49 5e 66 a4 be ba 0f 24 f2 59 dc

GDCA TrustAUTH E5 ROOT SHA1 digest = eb 46 6c d3 75 65 f9 3c de 10 62 cd 8d 98 26 ed 23 73 0f 12
GDCA TrustAUTH E4 EV SSL CA See “GDCA EV CPS”
GDCA TrustAUTH E4 OV SSL CA SHA1 digest = 50 15 62 d8 1b a2 40 27 1b ee 06 d2 b3 7f 5b 35 cb 9d 8c b8
GDCA TrustAUTH E4 IV SSL CA SHA1 digest = a8 45 2b fc 20 f9 de b6 9b 8b 3f 29 73 e0 a3 b3 6f 82 eb 5b
GDCA TrustAUTH E4 DV SSL CA SHA1 digest = 8e 9b 9a db f5 ec c4 6b 05 76 82 2e de 5e 80 d1 57 6b 5d 7c
GDCA TrustAUTH E4 CodeSigning CA SHA1 digest = 10 6a 4e 5d ca 05 92 28 e4 ff 89 52 66 53 a4 64 7d 57 ee 63
GDCA TrustAUTH E4 Generic CA SHA1 digest = fd 63 ba 6e e7 89 f6 0a 16 72 b5 b3 3a 29 7d 71 71 65 54 ee
GDCA TrustAUTH E4 Primer CA SHA1 digest =5f 42 a4 4d c8 ca 12 df ae 1c 29 92 1f 47 3e 3b be 8b d4 2c

There are also other changes:

Section 1.4.1.6. CP Object Identifiers of Certificates
Type I individual certificate policy: (1.2.156.112559.1.1.1.1)
Type II individual certificate policy: (1.2.156.112559.1.1.1.2)
Type III individual certificate policy: (1.2.156.112559.1.1.1.3)
Type IV individual certificate policy: (1.2.156.112559.1.1.1.4)
Type III organization certificate policy: (1.2.156.112559.1.1.2.1)
Type IV organization certificate policy: (1.2.156.112559.1.1.2.2)
Equipment certificate policy: (1.2.156.112559.1.1.3.1)
OV SSL server certificate policy: (1.2.156.112559.1.1.4.1)
IV SSL server certificate policy: (1.2.156.112559.1.1.4.2)
DV SSL server certificate policy: (1.2.156.112559.1.1.4.3)
EV SSL server certificate policy: (1.2.156.112559.1.1.6.1)
General CodeSigning certificate policy: (1.2.156.112559.1.1.5.1)
EV CodeSigning certificate policy: (1.2.156.112559.1.1.7.1)
Hong Kong-Guangdong mutual recognition individual certificates: 2.16.156.339.1.1.1.2.1
Hong Kong-Guangdong mutual recognition organization certificates: 2.16.156.339.1.1.2.2.1

Section 1.5.2. Contact Person
Contact: Ms Wang

[Han Yuwei]

在 2016年10月21日星期五 UTC+8下午12:15:07，wangs...@gmail.com写道：

I think this is a major mistake and a investgation should be conducted for CPS is a critical document about CA. This is not just a translation problem but a version control problem. Sometimes it can be lying.

[Jakob Bohm]

On 21/10/2016 10:38, Han Yuwei wrote:
>
> I think this is a major mistake and a investgation should be conducted for CPS is a critical document about CA. This is not just a translation problem but a version control problem. Sometimes it can be lying.
>

Let me try to be more specific:

When publishing a document called CPS version 4.3 the document with
that number must have the same contents in all languages that have a
document with that name and version number.

When making any change, even just correcting a mistyped URL, the
document becomes a new document version which should have a new and
larger number than the number of the document before the change.
Thus when a published document refers to a broken URL on your own
server, it is often cheaper to repair the server than to publish a new
document version. Some of the oldest CAs have been proudly
publishing their various important files at multiple URLs corresponding
to whatever was mentioned in old CP and CPS documents etc., only
shutting down those URLs years after the corresponding CA roots were
shut down.

There can also be a "draft" document which has no number and which
contains the changes that will go into the next numbered edition. Such
a "draft" would have no official significance, as it has not been
officially "published". For a well-planned change, the final "draft"
would be translated and checked into the relevant languages (e.g.
Chinese with mainland writing system, Chinese with Hong Kong and Macao
Special Administrative Regions old writing system, English), before
simultaneously publishing the matching documents with the same number
on the same day.

There are infinitely many version numbers in the universe to choose
from. There are also computer programs that can generate new version
numbers every time a draft is changed, but computers cannot decide when
a version is good enough in all languages to make an official
publication, and the computer generated version numbers are often
impractical for publication because they count all the small steps that
were not published.


Enjoy

Jakob
--
Jakob Bohm, CIO, Partner, WiseMo A/S. https://www.wisemo.com
Transformervej 29, 2860 Søborg, Denmark. Direct +45 31 13 16 10
This public discussion message is non-binding and may contain errors.
WiseMo - Remote Service Management for PCs, Phones and Embedded

[wangs...@gmail.com]

We have already implemented version control on Chinese version CP/CPS, which include version number (e.g. V4.3) and effective date (e.g. 2016-08-01). The revision and release of CP/CPS are reviewed and approved by the security policy committee (see section 1.5 in CP/CPS).

Meanwhile, we are a Chinese company and all of our customers are domestic. We only released the Chinese version of the CP/CPS, the English CP/CPS is only the translation version for reference.

It’s our negligence that we did a poor CP/CPS English translation for the latest version (CN version V4.3). After discussion and your suggestion, we will make sure that the same version of CP/CPS will be consistent in all languages in the future.

We've reconfirmed that the English version of CP/CPS uploaded by us in 2016-09-27 cannot be consistent with the latest Chinese version. We’ll upload the latest English version CP/CPS to you in this week.

Sorry for the inconvenience.

[Peter Kurrasch]

I think these are both good points and my recommendation is that Mozilla deny GDCA's request for inclusion.

We should not have to explain something as basic as document versioning and version control. If GDCA can not demonstrate sufficient controls over their documentation, there is no reason for the Internet community to place confidence in any of the other versioning systems that are needed to operate a CA.

Question: Are auditors expected to review translations of CP or CPS docs and verify consistency between them?

From: Jakob Bohm Sent: Saturday, October 22, 2016 9:07 AM To: mozilla-dev-s...@lists.mozilla.org Subject: Re: Guang Dong Certificate Authority (GDCA) root inclusion request

On 21/10/2016 10:38, Han Yuwei wrote:
>

> I think this is a major mistake and a investgation should be conducted for CPS is a critical document about CA. This is not just a translation problem but a version control problem. Sometimes it can be lying.
>

[wangs...@gmail.com]

在 2016年10月27日星期四 UTC+8上午8:09:06，Peter Kurrasch写道：

We’d like to explain a few points.

1. We have already implemented version control on Chinese version CP/CPS, the revision and release of CP/CPS are reviewed and approved by the security policy committee (see section 1.5 in CP/CPS). The Chinese version CP/CPS is also reviewed by our auditor.

2. The Chinese version CP/CPS is the formal documents we published in our Website. In the initial phase of "Bug 1128392", we have summited the Chinese version CP/CPS to Mozilla, and Mozilla release a basic review list in file "1128392-CAInformation.pdf" which contains instructions for us to summit some chapters of the CP/CPS in English version. We are not able to provide an accurate English version CP/CPS, but we will do our best to finish this translations and upload for reviewing process. We will upload the new English version CP/CPS for reference ASAP. However the English version CP/CPS should not be considered as formal documents. In case of any discrepancy between two versions, the Chinese version shall prevail.

3. Our auditor only reviews the Chinese version CP/CPS. It is not their responsibility to confirm the translated English versions.

[Han Yuwei]

在 2016年10月28日星期五 UTC+8上午2:12:32，Percy写道：

> According to Peter,

> "
> I reviewed the annual audit reports linked in your email, including
> the auditor's opinion and the management assertions.
>
> Good:
> - The reports and management assertion include an English language version
> - The English versions are authoritative (no qualification the Chinese
> language version holds in case of conflict)
> "
>

> This contradicts your assertion your assertion that

> "We are not able to provide an accurate English version CP/CPS, but we will do our best to finish this translations and upload for reviewing process. We will upload the new English version CP/CPS for reference ASAP. However the English version CP/CPS should not be considered as formal documents. In case of any discrepancy between two versions, the Chinese version shall prevail. "

Peter's meaning is about audit report itself.
I didn't see any language matter about CPS. The audit report said that its English version is authoritative not the CPS.
Maybe PricewaterhouseCoopers can explain which language did they audit.

[Han Yuwei]

在 2016年10月27日星期四 UTC+8下午6:22:03，wangs...@gmail.com写道：

Since Mozilla's working language is English (Not sure about this), it's your responsibility to provide an accurate translation of CPS. And I don't think the problem is poor translation. There is SO MANY DIFFERENCES between Chinese and English version. As a native Chinese speaker I don't believe the reason is poor translation but a bit more like you want to cover-up something.

What I most concern is about the CA certs and alogrithms. The English version didn't include serveral CA certs and SM2 and can't be explained by translation. For Wosign's SM2 cert accident, I suspect the missing of SM2 is intented.

What's more, there is another possibilty is that the English verison and Chinese version are maintained seperately.

[wangs...@gmail.com]

We are not intended to cover-up anything since we had disclosed every change to the Chinese version CP/CPS at once after the auditor reviewed.
The “ROOTCA(SM2)” CA in $1.1.3 of CPS ver4.3 is equivalent to the “SM2 ROOT Certificate” CA in $1.1.3 of CPS ver4.1. The “Guangdong Certificate Authority(SM2) ” CA in $1.1.3 of CPS ver4.3 is equivalent to the “SM2 CA Certificate” CA in $1.1.3 of CPS ver4.1. We change these names in diagram in this revision in order to show the actual CN of these certificates. Furthermore, we only issue SM2 subscriber certificates from the subCA of “ROOTCA(SM2)” CA.

[wangs...@gmail.com]

We have uploaded the lastest translantion of CP/CPS.
CP: https://bugzilla.mozilla.org/attachment.cgi?id=8805543
CPS: https://bug1128392.bmoattachments.org/attachment.cgi?id=8805545
EV CP: https://bugzilla.mozilla.org/attachment.cgi?id=8805546
EV CPS: https://bug1128392.bmoattachments.org/attachment.cgi?id=8805547

Because of our English level, there maybe some mistakes. If you have any questions, please contact us.

[Gervase Markham]

On 27/10/16 23:43, Han Yuwei wrote:
> Since Mozilla's working language is English (Not sure about this),

That is true.

> it's your responsibility to provide an accurate translation of CPS.

That is also true. However, we don't require that the English version be
the master copy.

Gerv

[Han Yuwei]

在 2016年10月28日星期五 UTC+8下午9:23:01，wangs...@gmail.com写道：

> We are not intended to cover-up anything since we had disclosed every change to the Chinese version CP/CPS at once after the auditor reviewed.
> The “ROOTCA(SM2)” CA in $1.1.3 of CPS ver4.3 is equivalent to the “SM2 ROOT Certificate” CA in $1.1.3 of CPS ver4.1. The “Guangdong Certificate Authority(SM2) ” CA in $1.1.3 of CPS ver4.3 is equivalent to the “SM2 CA Certificate” CA in $1.1.3 of CPS ver4.1. We change these names in diagram in this revision in order to show the actual CN of these certificates. Furthermore, we only issue SM2 subscriber certificates from the subCA of “ROOTCA(SM2)” CA.

Is SM2 acceptable in publicy-trusted CAs? I don't think so.

Maybe Gerv could explain more about this. And I am wondering what can CA do if government requirement conflicts with Mozilla's policy?

[Peter Bowen]

> On Oct 29, 2016, at 2:23 PM, Han Yuwei <hanyu...@gmail.com> wrote:
>
> 在 2016年10月28日星期五 UTC+8下午9:23:01，wangs...@gmail.com写道：

>> We are not intended to cover-up anything since we had disclosed every change to the Chinese version CP/CPS at once after the auditor reviewed.
>> The “ROOTCA(SM2)” CA in $1.1.3 of CPS ver4.3 is equivalent to the “SM2 ROOT Certificate” CA in $1.1.3 of CPS ver4.1. The “Guangdong Certificate Authority(SM2) ” CA in $1.1.3 of CPS ver4.3 is equivalent to the “SM2 CA Certificate” CA in $1.1.3 of CPS ver4.1. We change these names in diagram in this revision in order to show the actual CN of these certificates. Furthermore, we only issue SM2 subscriber certificates from the subCA of “ROOTCA(SM2)” CA.
>

> Is SM2 acceptable in publicy-trusted CAs? I don't think so.
>
> Maybe Gerv could explain more about this. And I am wondering what can CA do if government requirement conflicts with Mozilla's policy?

It is acceptable to have a single CPS that covers CAs that are included the Mozilla list of trust anchors and CAs that are not trusted by Mozilla. The CPS should make clear which portions apply to which CA when there are portions that do not apply to all CAs.

In this case, I would expect that the ROOTCA(SM2) CA is not proposed for inclusion in Mozilla. As long as the CPS does not allow issuance of SM2 signed certificates or certificates with SM2 subject public keys from the CAs proposed for inclusion in Mozilla, I do not seen an issue.

Thanks,
Peter

[Han Yuwei]

在 2016年10月28日星期五 UTC+8上午6:43:30，Han Yuwei写道：

To make my meaning more accurate, at this time I am NOT supporting the rejection to the inclusion request of GDCA. My proposes are (1.) to make GDCA follow the Mozilla's rules and response to our concerns in time (2.) and to make China Internet more secure. For now I am requesting GDCA to conduct a investgation about the discrepancy of CPS but the representive of GDCA insists on translation problems. If this is it, I am afraid that I would support the rejection.

P.S.
I do hope that there would be a Chinese CA which can obtain the trust of community but what I experienced now is disappointing me. Maybe I should wait for another decade.

[Han Yuwei]

在 2016年10月30日星期日 UTC+8上午5:30:23，Peter Bowen写道：

I don't see anything about this in Chinese CPS or Bugzilla. Could someone point out or GDCA explain about this?

[Gervase Markham]

On 29/10/16 22:23, Han Yuwei wrote:
> Is SM2 acceptable in publicy-trusted CAs? I don't think so.

No; the BRs list the permitted algorithms, and SM2 is not one of them.

> Maybe Gerv could explain more about this. And I am wondering what can
> CA do if government requirement conflicts with Mozilla's policy?

It may well be a government requirement that Chinese CAs be able to
issue SM2 certificates. However, no-one has yet demonstrated that it's a
requirement that they do so from specific roots (i.e. the ones trusted
by the major root stores).

Gerv

[wangs...@gmail.com]

在 2016年10月28日星期五 UTC+8上午8:19:43，Percy写道：
> "When facing any requirements of laws and regulations or any demands for undergoing legal
> process of court and other agencies, GDCA must provide confidential information in this CP"
>
> Can GDCA specify what other agencies are included? In China, many requests are relayed simply through a phone call without any paper trail or IM and the service providers must meet the demand very quickly. Are such informal procedures honored by GDCA?

Agencies include: public security organization, procuratorate, court.
The agency is required to meet the following conditions:
1. provide paper official letters
2. submit application to GDCA on site
3. site applicants must be law enforcement officers

[wangs...@gmail.com]

在 2016年10月30日星期日 UTC+8下午9:13:32，Gervase Markham写道：

We know that SM2 is not a permitted algorithm in BRs list. And we only apply for GDCA TrustAUTH R5 ROOT to be included this time.

[chinoc...@gmail.com]

wangs...@gmail.com於 2016年10月31日星期一 UTC+8下午2時22分05秒寫道：

I figured out that you are selling certificates on Taobao.com instead of handling application on your own website, does it matches your CP and CPS and if it conflicts with Mozilla's policy?

[Peter Bowen]

On Sun, Oct 30, 2016 at 11:34 PM, <chinoc...@gmail.com> wrote:
> wangs...@gmail.com於 2016年10月31日星期一 UTC+8下午2時22分05秒寫道：

> I figured out that you are selling certificates on Taobao.com instead of handling application on your own website, does it matches your CP and CPS and if it conflicts with Mozilla's policy?

There is nothing in Mozilla's policy that forbids CAs using resellers
and/or selling via 3rd party sites. Many, if not most, CAs in the
Mozilla program have reseller programs.

Thanks,
Peter

[Kathleen Wilson]

Thanks to all of you who have reviewed and commented on this request from Guangdong Certificate Authority (GDCA) is to include the "GDCA TrustAUTH R5 ROOT" certificate, turn on the Websites trust bit, and enabled EV treatment.

There were some recommendations to deny this request due to the versioning problems between the English documents and the original documents.

Do you all still feel that is the proper answer to this root inclusion request?

Or should we proceed with reviewing these new English translations of the documents, and make our decision based on the new versions?

Thanks,
Kathleen

[Han Yuwei]

在 2016年11月15日星期二 UTC+8上午8:51:25，Kathleen Wilson写道：

If GDCA insists on translation problem, I can't trust it.

[Gervase Markham]

On 15/11/16 08:39, Percy wrote:
> I posted on the solidot (Chinese Slashdot) about this. The majority
> comments want the application rejected.
> https://translate.google.com/translate?hl=en&sl=zh-CN&tl=en&u=http%3A%2F%2Fwww.solidot.org%2Fstory%3Fsid%3D50368

That fact, by itself, is not useful information. It would help if you
were to summarise why people feel the application should be rejected,
and how those reasons map to Mozilla's policy requirements.

Gerv

[wangs...@gmail.com]

在 2016年11月15日星期二 UTC+8上午8:51:25，Kathleen Wilson写道：

Because we misunderstand that we only need to provide the related chapters of CP/CPS in English, and non-related sections are not required. We are terribly sorry that we misinterpreted your requirement and upload an inconsistent CP/CPS in English. Someone inferred that we don’t utilize a version control for CP/CPS. In fact, we do have a strict control for master version CP/CPS (see section 1.5 in CP/CPS).
We understand that it is our responsibility to provide accurate English versions and ensure consistency and synchronicity between Chinese and English versions. Hence, we have decided to strictly implement version controlling of English version CP/CPS according to section 1.5 in CP/CPS. The auditor is reviewing our complete CP/CPS in English and the new version will be published as soon as possible.
We will keep open mind to process any further issues.

[Tony]

在 2016年11月15日星期二 UTC+8下午5:53:19，Gervase Markham写道：

Agreed with Gerv, the reasons for rejection/improvements should be listed very clearly in this case, considering the positive actions taken by CA as it will publish the full new version of CP/CPS in English and implement version control on EN CP/CPS.

It's common that Chinese company only maintains one master file policy which is in Chinese based on experience. But if they can do one version control on EN policies, even from now on, it should be deemed as a good improvement.

Also, guidance for coaching the CA to enhance the internal controls for the future is needed. More patience shall be given.

[谭晓生]

Agree with Gerv & Tony,
More patience should be given if they want to improve.

And I don’t think “I posted on the solidot (Chinese Slashdot) about this. The majority comments want the application rejected. “is enough to be the reason to reject the request.
For many Chinese companies, they do need to learn how to work with global community, they might even have language issues to use English as the working language, for Guang Dong CA case, I can see they are willing to work with the community and we should encourage them to do more.

Thanks，
Xiaosheng Tan



在 2016/11/15 下午9:08，“dev-security-policy 代表 Tony”<dev-security-policy-bounces+tanxiaosheng=360...@lists.mozilla.org 代表 wb10...@gmail.com> 写入:

在 2016年11月15日星期二 UTC+8下午5:53:19，Gervase Markham写道：

Agreed with Gerv, the reasons for rejection/improvements should be listed very clearly in this case, considering the positive actions taken by CA as it will publish the full new version of CP/CPS in English and implement version control on EN CP/CPS.

It's common that Chinese company only maintains one master file policy which is in Chinese based on experience. But if they can do one version control on EN policies, even from now on, it should be deemed as a good improvement.

Also, guidance for coaching the CA to enhance the internal controls for the future is needed. More patience shall be given.

[Han Yuwei]

在 2016年11月15日星期二 UTC+8下午7:03:07，wangs...@gmail.com写道：

Ok, this is what I want to see. Maybe next time you could be more specific about the problems and not just like "translation problem". If you can't describe your opinion exactly in English you can use Chinese and let others translate. But it's best for you to hire a professional translator.
Since CPS is very critical, I hope you understand what I said before. I don't want another Wosign incident happen again.

[Jakob Bohm]

Note that he said most of these things already in his post dated Thu,
27 Oct 2016 03:21:53 -0700 (PDT)

[Kevin]

Read through this discussion thread, here is my take on this. GDCA has taken action to address the discrepancies between English and Chinese CP/CPS and put in place stricter version control. So we should look at the updated version, if that's deemed good I don't see other reason to reject.

[Peter Bowen]

On Tue, Nov 15, 2016 at 3:02 AM, <wangs...@gmail.com> wrote:
>
> Because we misunderstand that we only need to provide the related chapters of CP/CPS in English, and non-related sections are not required. We are terribly sorry that we misinterpreted your requirement and upload an inconsistent CP/CPS in English. Someone inferred that we don’t utilize a version control for CP/CPS. In fact, we do have a strict control for master version CP/CPS (see section 1.5 in CP/CPS).
> We understand that it is our responsibility to provide accurate English versions and ensure consistency and synchronicity between Chinese and English versions. Hence, we have decided to strictly implement version controlling of English version CP/CPS according to section 1.5 in CP/CPS. The auditor is reviewing our complete CP/CPS in English and the new version will be published as soon as possible.
> We will keep open mind to process any further issues.

I think Mozilla needs to update its guidance to CAs. The information
checklist directions
(https://wiki.mozilla.org/CA:Information_checklist#Verification_Policies_and_Practices)
says "If the CP/CPS documents are not in English, then the portions of
those documents pertaining to verification of the certificate
subscriber must be translated into English."

This makes me think that the expectation is not that the full doc is
in English and that a one-time translation is acceptable.

I don't think we should hold it against GDCA that Mozilla's
requirements have apparently changed.

Thanks,
Peter

[Kathleen Wilson]

On Tuesday, November 15, 2016 at 10:41:28 AM UTC-8, Peter Bowen wrote:
> I think Mozilla needs to update its guidance to CAs. The information
> checklist directions
> (https://wiki.mozilla.org/CA:Information_checklist#Verification_Policies_and_Practices)
> says "If the CP/CPS documents are not in English, then the portions of
> those documents pertaining to verification of the certificate
> subscriber must be translated into English."

Done...
https://wiki.mozilla.org/CA:Information_checklist#Verification_Policies_and_Practices
now says: "If the CP/CPS documents are not in English, then the CP/CPS documents that are relevant to the root inclusion request must be translated into English."

Also updated
https://wiki.mozilla.org/CA:Recommended_Practices#Publicly_Available_CP_and_CPS
to add the second sentence to the 3rd bullet point:
"The CP/CPS should be available in an English version. The non-English version may be authoritative (as that's the working language of the CA) but the CA is responsible for ensuring that the translation is not materially different from the authoritative version of the document."

Note that this is also on our list of things to add directly to the policy:
https://github.com/mozilla/pkipolicy/issues/6

>
> This makes me think that the expectation is not that the full doc is
> in English and that a one-time translation is acceptable.
>
> I don't think we should hold it against GDCA that Mozilla's
> requirements have apparently changed.

Fair enough.

Before asking folks to review the documents again...

Would a representative of GDCA please confirm that the following translations are correct to the best of their knowledge?
Please also confirm that these documents match the corresponding version of the document in Chinese (no material differences).

CP: https://bugzilla.mozilla.org/attachment.cgi?id=8805543
CPS: https://bug1128392.bmoattachments.org/attachment.cgi?id=8805545
EV CP: https://bugzilla.mozilla.org/attachment.cgi?id=8805546
EV CPS: https://bug1128392.bmoattachments.org/attachment.cgi?id=8805547

Thanks,
Kathleen

[wangs...@gmail.com]

在 2016年11月16日星期三 UTC+8上午6:35:22，Kathleen Wilson写道：

Now we have a good understanding of the latest policy of Mozilla. We have sent a complete CP/CPS in English to our auditor. They will review the documents to ensure it is consistent with the Chinese version and meets the latest requirements. The CP/CPS in English will be revised and approved in light of Section 1.5 in CP/CPS after receiving feedback from the auditor, and will be published in our website before November 22nd 23:59 (Beijing time). We hope everyone can have a discussion based on the newly published versions.
Thanks to all for your understanding and suggestions to GDCA. We will keep an open mind to process any further issues.

[wangs...@gmail.com]

在 2016年11月16日星期三 UTC+8上午1:11:05，Han Yuwei写道：

Thanks for proposing many good questions, which push us to utilize version controls for English CP/CPS. We are looking forward to your further comments and suggestions.
We plan to attend the CA/B Forum meetings in February next year, If it is lucky to meet you there, we are looking forward to have your consultation and suggestions.

[Gervase Markham]

Hi Kathleen,

On 15/11/16 00:51, Kathleen Wilson wrote:
> There were some recommendations to deny this request due to the
> versioning problems between the English documents and the original
> documents.
>
> Do you all still feel that is the proper answer to this root
> inclusion request?

As I understand it, what happened was as follows:

* As part of their application, GDCA provided both Chinese and English
versions of their CP/CPS, posted to m.d.s.policy on 3rd August:

Chinese CP: http://www.gdca.com.cn/cp/cp
Chinese CPS: http://www.gdca.com.cn/cps/cps
English CP: https://bugzilla.mozilla.org/attachment.cgi?id=8650346
English CPS: https://bugzilla.mozilla.org/attachment.cgi?id=8688749

(I don't immediately have URLs for their EV CP and CPS in Chinese or
English from the original submission.)

* On 26th September, it was pointed out by Andrew Whalley that the
English versions had lower version numbers than the Chinese versions
(CP: 1.2 vs. 1.4; CPS: 4.1 vs 4.3)

* On 27th September, one day later, GDCA provided new English versions
with the same version numbers as the Chinese versions:

CP V1.4: https://bugzilla.mozilla.org/attachment.cgi?id=8795090
CPS V4.3: https://bugzilla.mozilla.org/attachment.cgi?id=8795091
EV CP V1.2: https://bugzilla.mozilla.org/attachment.cgi?id=8795093
EV CPS V1.3: https://bugzilla.mozilla.org/attachment.cgi?id=8795094

* It was pointed out by more than one person that there were significant
content differences between the English and Chinese versions which were
both labelled with the same version number

* GDCA said this was due to a "poor CP/CPS English translation" and on
28th October, provided new English versions (again) with the same
version numbers

CP: https://bugzilla.mozilla.org/attachment.cgi?id=8805543
CPS: https://bugzilla.mozilla.org/attachment.cgi?id=8805545
EV CP: https://bugzilla.mozilla.org/attachment.cgi?id=8805546
EV CPS: https://bugzilla.mozilla.org/attachment.cgi?id=8805547

What Mozilla has to decide is whether this was incompetence or malice.
Were GDCA trying to hide something? If so, their inclusion must be in
doubt. If they were not trying to hide something and just need a lesson
in version control, that is not necessarily something which
disqualifies, although it does give one concern.

Looking at the CPS (using pdf2txt and diff), the differences between the
originally-submitted v4.1 and the first 4.3 are very minor. One
intermediate certificate changes name throughout, as does the name of
GDCA. Three certs in an appendix are replaced with others. Other than
that, the only changes are these:

https://gist.github.com/gerv/fc311785c49c7fdfdfba78d6d5ad4aa9

This seems like an odd change, removing specificity about how domain
validation is done. This change was _added_ to the Chinese version of
3.2.5 between 4.1 and 4.2, and moved to section 3.2.7 in version 4.3. So
how does going from 4.1 to 4.3 in the English version lead to it being
removed?

The differences between the first 4.3 and the second one are much more
extensive.

So I'd say the questions for GDCA are these:

* When you were asked to produce a version of your CPS matching Chinese
version 4.3, within a day you came up with:
https://bugzilla.mozilla.org/attachment.cgi?id=8795091
That clearly doesn't match Chinese version 4.3, and yet it has "version
4.3" written in it. And the effective date marked within it is one month
_earlier_ than the effective date of the Chinese 4.3. How did this
happen? How did such a document come to exist with such a version number
and date attached, when it is so massively different from the real 4.3,
and so similar to the previous 4.1?

* You say you only translated the relevant bits rather than all of it,
which is why there is a discrepancy, but the diff between 4.1 and the
first version of 4.3 reveals no additions, only one deletion. How does
that fit?

Gerv

[Han Yuwei]

在 2016年11月16日星期三 UTC+8下午3:59:12，wangs...@gmail.com写道：

So are you going to conduct a complete investgation about this? I am still waiting for it.

[Jakob Bohm]

I believe their overall excuses can be rephrased as:

1. The previous Mozilla policy guidance only requiring a partial
translation. Mozilla has since fixed that, though I can't find
the posting that mentioned that document fix.

2. Sloppy translation work, including sloppyness when trying to quickly
update the 4.1 translation to match the 4.3 Chinese documents.

3. At the time, the English translations were not version controlled,
only the Chinese versions. They have promised to change this for the
next version, where they will even request an outside audit of the
translation.

4. That consistent pair of a new CP/CPS in Chinese and English has not
been posted yet, indicating that Mozilla will just have to put the
inclusion request on hold until then.

[wangs...@gmail.com]

在 2016年11月17日星期四 UTC+8下午7:20:05，Gervase Markham写道：

On 2015.8.20, we uploaded the English Version 4.1 to bugzilla for the first time, and section 3.2.5 was translated from the Chinese version CPS 4.1.
On 2015.10.22, Kathleen pointed out that section 3.2.5 " Domain name recognition and identification" could not meet the requirements of Mozilla.
Therefore, on 2015.11.17, we submitted a new English Version 4.1 with modified section 3.2.5.
However, due to the negligence of our employee, the submitted English Version 4.3 on 2016.9.26 was revised based on the English version 4.1 of 2015.8.20.
In the submitted English Version 4.3 on 2016.10.28, this mistake was found and solved.
For the question that "the Chinese version of 3.2.5 between 4.1 and 4.2 moved to section 3.2.7 in version 4.3", this is because we added section 3.2.5 " Authentication of SSL Server Identity" and section 3.2.6 " Authentication of CodeSigning Identity". Accordingly, the former 3.2.5 changes into 3.2.7.

>
> The differences between the first 4.3 and the second one are much more
> extensive.
>
> So I'd say the questions for GDCA are these:
>
> * When you were asked to produce a version of your CPS matching Chinese
> version 4.3, within a day you came up with:
> https://bugzilla.mozilla.org/attachment.cgi?id=8795091
> That clearly doesn't match Chinese version 4.3, and yet it has "version
> 4.3" written in it. And the effective date marked within it is one month
> _earlier_ than the effective date of the Chinese 4.3. How did this
> happen? How did such a document come to exist with such a version number
> and date attached, when it is so massively different from the real 4.3,
> and so similar to the previous 4.1?
>
> * You say you only translated the relevant bits rather than all of it,
> which is why there is a discrepancy, but the diff between 4.1 and the
> first version of 4.3 reveals no additions, only one deletion. How does
> that fit?

GDCA takes security and governance seriously and we have a strict control for Chinese version CP/CPS, all the contents are disclosed. And The Chinese versions for CPS 4.1, 4.2, 4.3 are published on the official website, so we cannot cover-up anything. There are three reasons for the confusion in the English versions: 1) The English versions don't have a version control or audit process and we misunderstand that we only need to provide the related chapters of CP/CPS in English; 2) For fear of affecting the progress of inclusion, we were eager to upload new English versions quickly , which resulting in some omissions and inconsistent etc. 3) During the translation period, our company name changed, some new root certificates and intermediate certificates are issued, point/time of period audit were made and so on, so the Chinese versions were updated several times, but the employee who is responsible for the translation didn't translated into English version timely. Now we strictly implement version controlling of English version CP/CPS, and have sent the English versions to the auditor for review, the new versions will be published before next Tuesday 23:59.

[Gervase Markham]

On 18/11/16 11:38, wangs...@gmail.com wrote:
> GDCA takes security and governance seriously and we have a strict
> control for Chinese version CP/CPS, all the contents are disclosed.
> And The Chinese versions for CPS 4.1, 4.2, 4.3 are published on the
> official website, so we cannot cover-up anything. There are three
> reasons for the confusion in the English versions:

<snip>

Thank you for this information. I accept these explanations. Once the
fully-translated documents are available, I think the inclusion request
should continue.

Gerv

[wangs...@gmail.com]

在 2016年11月18日星期五 UTC+8下午10:05:46，Gervase Markham写道：

We appreciate for your understanding and suggestion. Let's work together to make the Internet a safer and more trusted space for everyone.

[wangs...@gmail.com]

Thanks for all suggestions upon our CP/CPS and base on the development of our business, we have revised and prepared a bilingual edition of CP/CPS, which have been submitted to our auditor to check the consistency of major contents between Chinese version and English version, and officially published after internal review and approval in accordance with control regulations disclosed in CP. We did the translation work to the best of our knowledge. However, there may still be some improper translation because of time limit. Please kindly point out if there is any error, propose any question or offer suggestions for us to make further improvements in the next revision. We will devote our best effort to do it better!
The following files have been published in the official website of GDCA and will come into effect on Dec 1st, 2016. Here are the names of the files and their URL:
1. GDCA CP Ver 1.5
https://www.gdca.com.cn/export/sites/default/customer_service/.content/attachments/1.GDCA-CP-V1.5.pdf
2. GDCA CPS Ver 4.4
https://www.gdca.com.cn/export/sites/default/customer_service/.content/attachments/1.GDCA-CPS-V4.4.pdf
3. GDCA EV CP Ver 1.3
https://www.gdca.com.cn/export/sites/default/customer_service/.content/attachments/1.GDCA-EVCP-V1.3.pdf
4. GDCA EV CPS Ver 1.4
https://www.gdca.com.cn/export/sites/default/customer_service/.content/attachments/1.GDCA-EVCPS-V1.4.pdf

[Gervase Markham]

On 22/11/16 15:47, wangs...@gmail.com wrote:
> into effect on Dec 1st, 2016. Here are the names of the files and
> their URL: 1. GDCA CP Ver 1.5
> https://www.gdca.com.cn/export/sites/default/customer_service/.content/attachments/1.GDCA-CP-V1.5.pdf

A bilingual edition seems like an excellent idea :-)

I would note that the site from which you are serving these documents
uses a security certificate which is not currently trusted by Firefox.
Is that because it uses the new root you are applying to include?

Gerv

[wangs...@gmail.com]

The answer is yes. That’s why we need to apply for root inclusion. We also upload the latest version of CP/CPS here for your convenience.

1. GDCA CP Ver 1.5

https://bug1128392.bmoattachments.org/attachment.cgi?id=8813656

2. GDCA CPS Ver 4.4

https://bug1128392.bmoattachments.org/attachment.cgi?id=8813658

3. GDCA EV CP Ver 1.3

https://bug1128392.bmoattachments.org/attachment.cgi?id=8813659

4. GDCA EV CPS Ver 1.4

https://bug1128392.bmoattachments.org/attachment.cgi?id=8813660

[Kathleen Wilson]

I will greatly appreciate it if folks will review and comment on these updated CP/CPS documents.

Thanks,
Kathleen

[wangs...@gmail.com]

We compiled an analysis document on our CP/CPS’s Compliance with the BRs for everyone to review and comment. You can find the document at the following address of the BUG:https://bug1128392.bmoattachments.org/attachment.cgi?id=8851230
 
Your suggestions will be much appreciated.

[Kathleen Wilson]

All,

This request is to include the "GDCA TrustAUTH R5 ROOT" certificate, turn on the Websites trust bit, and enabled EV treatment.

In order to help get this discussion moving again, I asked GDCA to provide a side-by-side comparison of the latest version of the BRs with their CP/CPS documents.

They provided this BR-self-assessment here:
https://bugzilla.mozilla.org/attachment.cgi?id=8851230

The documents that were evaluated in this self-assessment are available on the CA's website.

All of these documents contain both the original text in Chinese, and the translation into English.

Document Repository:
https://www.gdca.com.cn/customer_service/knowledge_universe/cp_cps/

GDCA CP v1.5:
https://www.gdca.com.cn/customer_service/knowledge_universe/cp_cps/CPCPS-GDCA1.5GDCA-CP-V1.5/

GDCA CPS v4.4:
https://www.gdca.com.cn/customer_service/knowledge_universe/cp_cps/CPCPS-GDCA4.4GDCA-CPS-V4.4/

GDCA EV CP v1.3:
https://www.gdca.com.cn/customer_service/knowledge_universe/cp_cps/CPCPS-GDCA-EV1.3GDCA-EV-CP-V1.3/

GDCA EV CPS v1.4:
https://www.gdca.com.cn/customer_service/knowledge_universe/cp_cps/CPCPS-GDCA-EV1.4GDCA-EV-CPS-V1.4/

I will greatly appreciate it if you all would take another look at this CA's request, review their self-assessment, and respond in this discussion to let me know if you believe that this CA has addressed all of your questions or concerns.

Also, I would like to make this BR-self-assessment a standard part of Mozilla's root inclusion/change process. I will draft what that will look like, and start a separate discussion about it.

Thanks,
Kathleen

[Patrick Tronnier]

On Sunday, March 26, 2017 at 11:48:43 PM UTC-4, wangs...@gmail.com wrote:
> We compiled an analysis document on our CP/CPS’s Compliance with the BRs for everyone to review and comment. You can find the document at the following address of the BUG:https://bug1128392.bmoattachments.org/attachment.cgi?id=8851230
>  
> Your suggestions will be much appreciated.

As part of the suggestion process it would be useful to expand on the tables you listed in section 2 "Compliance Analysis". Would you be able to attach an editable MS Word version?

[wangs...@gmail.com]

在 2017年3月30日星期四 UTC+8下午10:34:00，Patrick Tronnier写道：

We uploaded an editable version of the comparison document in MS Word format for the convenience of review and comments. You can find the document at the following address of the BUG:https://bug1128392.bmoattachments.org/attachment.cgi?id=8853351
The contents of the Word document are the same with that of the PDF document, except for the correction of a mistake in the Table of Contents part of the PDF document. All suggestions and comments are welcomed. Thanks.

[wangs...@gmail.com]

We have just published the updated CP/CPS documents, this version has been revised according to the latest Baseline Requirements and has been reviewed internally, meanwhile, the points our “Analysis on the Compliance of GDCA’s CP and CPS with the Baseline Requirements (published on March 25, 2017)” promised to disclose have been included in this version, and we will update the compliance analysis document as soon as possible. Please find the new version at:
CP V1.6: https://bug1128392.bmoattachments.org/attachment.cgi?id=8860016
CPS V4.5: https://bug1128392.bmoattachments.org/attachment.cgi?id=8860018
EV CP V1.4: https://bug1128392.bmoattachments.org/attachment.cgi?id=8860019
EV CPS V1.5: https://bug1128392.bmoattachments.org/attachment.cgi?id=8860020

We wish these documents will be fully discussed by the public, so that Mozilla can make decision on this root inclusion application.
All comments and suggestions are welcomed. Thanks.

[Patrick Tronnier]

I updated your bug with a review of your initial BR-self-assessment using the previously posted CPS's. The review is attachment https://bugzilla.mozilla.org/attachment.cgi?id=8860075.

Would you please complete a second BR-self-assessment against the just posted CPS's and CP's and use my attachment as your starting point? Thank you.

[wangs...@gmail.com]

在 2017年4月20日星期四 UTC+8下午11:31:14，Patrick Tronnier写道：

Hi Patrick,

Thanks for the comments.

Please check our second BR self-assessment against our updated CP/CPS (CP V1.6, CPS V4.5, EV CP V1.4, and EV CPS V1.5).You can find the document at the following address of the BUG:https://bugzilla.mozilla.org/attachment.cgi?id=8860627

We welcome all comments and suggestions.
Thanks.

[Andrew R. Whalley]

Greetings,

I've given the CP V1.6 and CPS V4.5 docs a quick looking through and have
the following comments:

* Please don't protect your PDFs for printing

* https://SSLTEST-2.95105813.cn - which I believe should be revoked, has
also expired. The revoked test cert should be otherwise valid and not
expired.

* While there is mention of how CAA records are dealt with in section
4.2.4, it doesn't seem to specify what value is expected to be present in
the record for the check to pass.

* 3.2.7. 域名的确认和鉴别 Domain name recognition and identification - This section
references BR version v1.4.1. Version 1.4.4 is current and has changes in
the section numbers referred to here. (Also see versions under IPR review
on the CAB Forum website)

* 7.1 Certificate profile: There is no mention of how the serial number is
generated. The BRs specify "Effective September 30, 2016, CAs SHALL
generate non‐sequential Certificate serial numbers greater than zero (0)
containing at least 64 bits of output from a CSPRNG."

* CP 7.1.3 says "The cryptographic algorithm identifiers of certificates
issued by GDCA include sha1RSA, sha256RSA and sha256ECDSA". SHA1
signatures must not be used in any part of the publically trusted hierarchy.

* CP 7.1.5 on "Name Constraints" looks like it's referring to 3.1.2 "Need
for names to be meaningful". This section is meant to refer to RFC 5280
section 4.2.1.10 name constraints.

* CPS Appendix1: Certificate information of the publicly trusted CAs: Most
of the listed CAs can't be found in crt.sh - it would be great to get them
CT logged.

* Only GDCA TrustAUTH R5 ROOT (SHA-1 Fingerprint
0F:36:38:5B:81:1A:25:C3:9B:31:4E:83:CA:E9:34:66:70:CC:74:B4) seems to have
been disclosed in Mozilla's CCADB.

Thanks,

Andrew

On Sat, Apr 22, 2017 at 5:08 AM, wangsn1206--- via dev-security-policy <
dev-secur...@lists.mozilla.org> wrote:

> 在 2017年4月20日星期四 UTC+8下午11:31:14，Patrick Tronnier写道：

> Hi Patrick,
>
> Thanks for the comments.
>
> Please check our second BR self-assessment against our updated CP/CPS (CP

> V1.6, CPS V4.5, EV CP V1.4, and EV CPS V1.5).You can find the document at
> the following address of the BUG:https://bugzilla.mozilla.

> org/attachment.cgi?id=8860627
>
> We welcome all comments and suggestions.
> Thanks.

> _______________________________________________
> dev-security-policy mailing list
> dev-secur...@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-security-policy
>

[wangs...@gmail.com]

Hi Andrew,

Thanks for the comments. Please check our following responses.

> * Please don't protect your PDFs for printing
>

We have removed the restrictions on the printing of the PDF documents and re-uploaded them to the BUG, these documents are available at:

https://bug1128392.bmoattachments.org/attachment.cgi?id=8866668
https://bug1128392.bmoattachments.org/attachment.cgi?id=8866669
https://bug1128392.bmoattachments.org/attachment.cgi?id=8866670
https://bug1128392.bmoattachments.org/attachment.cgi?id=8866671

> * https://SSLTEST-2.95105813.cn - which I believe should be revoked, has
> also expired. The revoked test cert should be otherwise valid and not
> expired.
>

We have replaced the certificate. Please check.

> * While there is mention of how CAA records are dealt with in section
> 4.2.4, it doesn't seem to specify what value is expected to be present in
> the record for the check to pass.
>

GDCA will not issue corresponding certificates if the "issue", "issuewild"property tags do not contain“gdca.com.cn”. In case the property tag “iodef” is present in the CAA records, GDCA will determine whether or not to issue certificates after communicating with the applicant.

> * 3.2.7. 域名的确认和鉴别 Domain name recognition and identification - This section
> references BR version v1.4.1. Version 1.4.4 is current and has changes in
> the section numbers referred to here. (Also see versions under IPR review
> on the CAB Forum website)
>

As required by Mozilla, CPS V4.5 uses methods documented in section 3.2.2.4 of version 1.4.1 of the CA/Browser Forum's Baseline Requirements (BRs)for domain validation. We will checkBR v1.4.3 and later versions and update this section as necessary.

> * 7.1 Certificate profile: There is no mention of how the serial number is
> generated. The BRs specify "Effective September 30, 2016, CAs SHALL
> generate non‐sequential Certificate serial numbers greater than zero (0)
> containing at least 64 bits of output from a CSPRNG."
>

In our next version, we will disclose: “Certificate serial number--Within the domain of each Issuing CA, GDCA includes a unique non-sequential Certificate serial number greater than zerocontaining at least 64 bits of output from a CSPRNG.”

> * CP 7.1.3 says "The cryptographic algorithm identifiers of certificates
> issued by GDCA include sha1RSA, sha256RSA and sha256ECDSA". SHA1
> signatures must not be used in any part of the publically trusted hierarchy.
>

Certificates of GDCA's publically trusted CAs do not use sha1RSA signing algorithm. We will definitely disclose in the next version.

> * CP 7.1.5 on "Name Constraints" looks like it's referring to 3.1.2 "Need
> for names to be meaningful". This section is meant to refer to RFC 5280
> section 4.2.1.10 name constraints.
>

The content of section 7.1.5 of CP/CPS will be moved to section 3.1.2 in the next version of CP/CPS. And GDCA has no stipulation on Name Constraints.

> * CPS Appendix1: Certificate information of the publicly trusted CAs: Most
> of the listed CAs can't be found in crt.sh - it would be great to get them
> CT logged.
>

Already get CT logged for GDCA TrustAUTH R5 ROOT,and such operation cannot be done for the other two ROOTs as we have not yet applied for the inclusion of these two.

> * Only GDCA TrustAUTH R5 ROOT (SHA-1 Fingerprint
> 0F:36:38:5B:81:1A:25:C3:9B:31:4E:83:CA:E9:34:66:70:CC:74:B4) seems to have
> been disclosed in Mozilla's CCADB.
>

We have uploaded all the intermediate certificates of the GDCA TrustAUTH R5 ROOT in the CCADB. According to the CCADB rules, CAs can only upload intermediate certificates, and only Root Store Members can upload root certificates, therefore, GDCA has not uploaded the 数安时代R5根CA and the GDCA TrustAUTH E5 ROOTand their respective intermediate certificates.

For your information, we expect to publish the next version of our CP/CPS on around May 25, 2017, with the latest comments and BR v1.4.4 taken into consideration.

And as always, we welcome all comments and suggestions.

Thanks.

[Rob Stradling]

On 11/05/17 10:42, wangsn1206--- via dev-security-policy wrote:
<snip>

>> * CPS Appendix1: Certificate information of the publicly trusted CAs: Most
>> of the listed CAs can't be found in crt.sh - it would be great to get them
>> CT logged.
>>

> Already get CT logged for GDCA TrustAUTH R5 ROOT,and such operation cannot be done for the other two ROOTs as we have not yet applied for the inclusion of these two.

Hi.

Would you like me to add your other two roots to the list of roots that
are accepted by the Comodo Dodo log?

If so, please either submit a pull request on GitHub (see
https://github.com/Comodo-CA/CTLogs-AcceptedRoots) or send me the two
root certs.

The Comodo Dodo log isn't trusted by Chrome, but it is monitored by crt.sh.

--
Rob Stradling
Senior Research & Development Scientist
COMODO - Creating Trust Online

[wangs...@gmail.com]

在 2017年5月11日星期四 UTC+8下午5:58:00，Rob Stradling写道：

Hi Rob,

Thanks for your kind assistance, we have e-mailed our roots to you for this purpose.

[Patrick Tronnier]

Greetings, I have reviewed your second BR self-assessment (https://bugzilla.mozilla.org/attachment.cgi?id=8860627) against your updated CP/CPS (CP V1.6, CPS V4.5, EV CP V1.4, and EV CPS V1.5) and provided the following comments and/or recommendations.


1. BR Section 3.2.2.5 Authentication for an IP: Per your comments please make sure your CPS states “GDCA does not issue EV certificate for an IP address.”

2. BR Section 3.2.2.7 Data Source Accuracy: I recommend adding the specific length of time data is relied upon (i.e. 39 months or 825 days per BRs) to section 3.2.11 of your CPS.

3. BR Section 3.2.2.7 Data Source Accuracy: I recommend adding the specific length of time data is relied upon (i.e. 39 months or 825 days per BRs) to section 3.2.7 of your EV CPS.

4. BR Section 3.2.3 Authentication of Individual Identity: I do not see in the CPS/CP where the differences in authentication of individuals is backed up by the appropriate technical constraining of the type of certificate issued.
4.1. Your comments for Type I and Type II Individual Certificates state they “are only for ordinary signing certificates, not for SSL certificates and code signing certificates” but I can’t find in the CPS where this is substantiated. I recommend clearly documenting in the CPS how each type of certificate is technically constrained (i.e. Key Usage, Enhanced Key Usage, etc.) and in CPS section 1.3.7.1 removing the words “but not limited to”.
4.2. For Type III certificates change the word “can” to “must”. (i.e. This must be validated by ID card, officer card or other valid document issued by government agency.”

5. BR Section 3.2.5 Validation of Authority: Per your comments please make sure this is clearly defined in the next version of your CPS.

6. BR Section 3.2.6 Criteria for Interoperation or Certification. Per your comments please make sure the next version of your CPS states you do not issue any cross certificates.

7. BR Section 4.2.1 Performing Identification and Authentication Functions. Per your comments please make sure the next version of your CPS states you do not rely on data older than 27 months (or 39 months or 825 days per BRs).

8. BR Section 4.2.2 Approval or Rejection of Certificate Applications: Per your comments please make sure the next version of your CPS states GDCA does not issue certificates containing a new gTLD under consideration by ICANN.

9. BR Section 4.3.1 CA Actions during Certificate Issuance: Per your comments please make sure the next version of your CPS states “Certificate issuance by the Root CA SHALL require an individual authorized by the CA (i.e. the CA system operator, system officer, or PKI administrator) to deliberately issue a direct command in order for the Root CA to perform a certificate signing operation.”

10. BR Section 4.5.1 Subscriber private key and certificate usage: Per your comments please make sure the next version of your CPS details the use of SSL certificates per #4 (Use of Certificate) as described in BR Section 9.6.3. Subscriber Representations and Warranties.

11. BR Section 4.9.13 Circumstances for Suspension: Per your comments please make sure the next version of your CPS states certificate suspension is not allowed.

12. BR Section 4.10.1 Operational Characteristics: Per your comments please make sure the next version of your CPS states “Revocation entries on a CRL or OCSP Response will not be removed until after the Expiry Date of the revoked Certificate”.

13. BR Section 4.10.2 Service Availability: Per your comments please make sure the next version of your CPS states “the service response time shall be less than 10 seconds”.

14. Based on your self assessment comments in BR sections 1 – 4, I submit it would be useful for you to revisit your assessment of BR sections 5 (MANAGEMENT, OPERATIONAL, AND PHYSICAL CONTROLS) through section 9 (OTHER BUSINESS AND LEGAL MATTERS) and update your BR Assessment.

[wangs...@gmail.com]

在 2017年5月16日星期二 UTC+8上午5:19:14，Patrick Tronnier写道：

Hi Patrick, 

Thanks for the comments and recommendations, which will be duly considered as necessary in the next version of our CP/CPS to be published on around May 25, 2017.
Thanks. 

[Rob Stradling]

On 12/05/17 06:51, wangsn1206--- via dev-security-policy wrote:
> 在 2017年5月11日星期四 UTC+8下午5:58:00，Rob Stradling写道：

>> On 11/05/17 10:42, wangsn1206--- via dev-security-policy wrote:
>> <snip>
>>>> * CPS Appendix1: Certificate information of the publicly trusted CAs: Most
>>>> of the listed CAs can't be found in crt.sh - it would be great to get them
>>>> CT logged.
>>>>
>>> Already get CT logged for GDCA TrustAUTH R5 ROOT,and such operation cannot be done for the other two ROOTs as we have not yet applied for the inclusion of these two.
>>
>> Hi.
>>
>> Would you like me to add your other two roots to the list of roots that
>> are accepted by the Comodo Dodo log?
>>
>> If so, please either submit a pull request on GitHub (see
>> https://github.com/Comodo-CA/CTLogs-AcceptedRoots) or send me the two
>> root certs.
>>
>> The Comodo Dodo log isn't trusted by Chrome, but it is monitored by crt.sh.
>

> Hi Rob,
>
> Thanks for your kind assistance, we have e-mailed our roots to you for this purpose.

Our Dodo log now accepts the "GDCA TrustAUTH E5 ROOT" and "数安时代 R5 根 CA"
root certificates, and I've submitted both of them to Dodo. You can see
them here:

https://crt.sh/?id=139646527
https://crt.sh/?id=139646529

[wangs...@gmail.com]

在 2017年5月17日星期三 UTC+8下午5:18:59，Rob Stradling写道：

Thanks Rob.

[wangs...@gmail.com]

Hi All,

We have just updated and published our CP/CPS, and the latest versions are available at:
CP V1.7: https://bug1128392.bmoattachments.org/attachment.cgi?id=8871236
CPS V4.6: https://bug1128392.bmoattachments.org/attachment.cgi?id=8871237
EV CP V1.5: https://bug1128392.bmoattachments.org/attachment.cgi?id=8871238
EV CPS V1.6: https://bug1128392.bmoattachments.org/attachment.cgi?id=8871240

We wish these documents will be fully discussed and commented.
Thanks.

[Kathleen Wilson]

The updated documents are also posted on the CA's website:
https://www.gdca.com.cn/customer_service/knowledge_universe/cp_cps/

Current audit statements are here:
WebTrust CA: https://cert.webtrust.org/ViewSeal?id=2231
WebTrust BR: https://cert.webtrust.org/ViewSeal?id=2232
WebTrust EV SSL: https://cert.webtrust.org/ViewSeal?id=2233

Unless anyone raises further questions or concerns, I plan to close this discussion and recommend approval of this request to include "GDCA TrustAUTH R5 ROOT" certificate, turn on the Websites trust bit, and enabled EV treatment.

Thanks,
Kathleen

[Kathleen Wilson]

Thanks to all of you who reviewed and commented on this request from Guangdong Certificate Authority (GDCA) to include the GDCA TrustAUTH R5 ROOT certificate, turn on the Websites trust bit, and enabled EV treatment.

I believe that all of the concerns that were raised in this discussion have been properly addressed, and I will state my intent to approve this request in the bug.

https://bugzilla.mozilla.org/show_bug.cgi?id=1128392

I am now closing this discussion. Any further follow-up should be added directly to the bug.

Thanks,
Kathleen