1579060 - mozilla::pkix tag definitions for issuerUniqueID and subjectUniqueID shouldn't have the CONSTRUCTED bit set

Status: RESOLVED FIXED

(NSS :: Libraries, defect, P1)

Description

[Stefan Matthäus]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:69.0) Gecko/20100101 Firefox/69.0

Steps to reproduce:

I open management https websites from server hardware like DELL iDRAC, VMWare ESXi, Webmin, etc. which only have self signed certificate with Firefox 69. (OS = Windows 10 1903)

Actual results:

Firefox displays:

Fehler: Gesicherte Verbindung fehlgeschlagen

Beim Verbinden mit idrac.server.fqdn trat ein Fehler auf. Sicherheitsbibliothek: Fehlerhaft formatierte DER-verschlüsselte Nachricht. Fehlercode: SEC_ERROR_BAD_DER

(...bla...)
/quote

When clicking on (i) in front of the URL, it says connection is not safe (what is true due to self signed certificate on that device) , then click on ">" & more information, going to the security tab, it tells that the website has no owner and the website is not encrypted.

Expected results:

Open the website, like it still happened with Firefox 68.0.2 before.

Comment:

I can understand, that self signed websites are not the acme of security. But does it really make sense to block them totally? Then it even it is no more possible to secure them on device installation/maintenance, as you can't get into the device's management interface to assign a secure certificate. This is an egg-chicken-problem and it would be required that the user can override such blocking like it has been before. Otherwise user can not secure the device through it's web based management interface.

Or, isn't it a feature, but it is a bug?

Comment 1

[BugBot [:suhaib / :marco/ :calixte]]

Bugbug thinks this bug should belong to this component, but please revert this change in case of error.

Comment 2

[J.C. Jones [:jcj] (he/they)]

This wouldn't be a self-signed issue, it sounds like the cert itself has an encoding problem.

Any chance you can attach the certificate to the bug? I am not sure what ASN.1 decoding changes we may have made that might have caused this, but that's how we're going to nail it down.

Thanks!

Comment 3

[Stefan Matthäus]

Hello, how to attach screenshots from firefox, chrome and edge? Chrome&Edge warns because of self signed cert but I can overide and access. The website example is a Dell iDRAC server management website.

Base 64 encoded self signed certificate of the Dell iDRAC (saved using Chrome browser, Firefox does not offer as it does not detect the cerificate at all)

-----BEGIN CERTIFICATE-----
MIIEETCCAvmgAwIBAgIJAK/M8I2XrH5vMA0GCSqGSIb3DQEBCwUAMIGdMQswCQYD
VQQGEwJVUzEOMAwGA1UECAwFVGV4YXMxEzARBgNVBAcMClJvdW5kIFJvY2sxEjAQ
BgNVBAoMCURlbGwgSW5jLjEcMBoGA1UECwwTUmVtb3RlIEFjY2VzcyBHcm91cDEW
MBQGA1UEAwwNaWRyYWMtSjM4UTNLMjEfMB0GCSqGSIb3DQEJARYQc3VwcG9ydEBk
ZWxsLmNvbTAgGA8yMDE3MDUxMDIyNTQwMFoXDTI3MDUxMTIyNTQ1MlowgZ0xCzAJ
BgNVBAYTAlVTMQ4wDAYDVQQIDAVUZXhhczETMBEGA1UEBwwKUm91bmQgUm9jazES
MBAGA1UECgwJRGVsbCBJbmMuMRwwGgYDVQQLDBNSZW1vdGUgQWNjZXNzIEdyb3Vw
MRYwFAYDVQQDDA1pZHJhYy1KMzhRM0syMR8wHQYJKoZIhvcNAQkBFhBzdXBwb3J0
QGRlbGwuY29tMIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAp08QAk1T
SnHLECO0n8AxPA98OjqPZDF9Xu7qKE5ewbGLG7n1vbU/nT0qQyOmXxopj4SGfpQh
llJwGRPqlcTNHPVBwWnPWwOU5LC1up4KLF7qt6y5wnd8uJ/v/3+GkjIEz6SvbP9m
P62SpT3lmgB+yaCeWGoXrYFkqzfi+WPv6Y/DEA3sThEJokfUSgRApwMZW41autY0
fw1ZpY6XhSRd8AdpZ6eJ8BtTZpgafxEtgcvJb+Iv2XU2+55S4LHlzR7YtHl8KZoa
v3q6gelglg6JL+rKAZvboGMzqlQRgsy0mbLyaBUhbkjN9+Rpeuxg+tsSQ5RznhzN
POVvsP4qD+9t7QIDAQABo1AwTjAdBgNVHQ4EFgQUhJvVFuSlkBHQZ3FZN3txo4C6
AY0wHwYDVR0jBBgwFoAUhJvVFuSlkBHQZ3FZN3txo4C6AY0wDAYDVR0TBAUwAwEB
/zANBgkqhkiG9w0BAQsFAAOCAQEAlr929L+EHeqdZdWinjdkw3HYVso9J+D9ArNq
9RZk1+t+NVCa77RFvXPk+h1mB/sKs4REYFeNw7Udo9VIHKUz5Xt+ALI+q9r+qkeF
SR2o0168zoZHvE66XATZTUmlVL84sZ+QsUapG/PRneeGSy1XH3GB1QheimtkQdBN
sNWrxDAhSuukEuBGPNoSUzgRgihDxIMS/SMK+S264hTM18EosZD/QFmnaohsv50p
CDVTrEP75U+hJTAOvP6KjFT2we0YZCP9RgGJcLstLFYiUS1WAe5JO2DEf2fcfOnl
xEswB9XYD5eBAuMPfzRuBz1FK3XBW0yhHNBMMafcy1PnKINGkQ==
-----END CERTIFICATE-----

Comment 4

[Dana Keeler (she/her) (use needinfo) [:keeler]]

Can you also attach a packet trace of the TLS handshake? Thanks!

Comment 5

[Cédric]

Hi,
There are people that confirm the problem on reddit: https://www.reddit.com/r/firefox/comments/d50gv7/firefox_ver69_blocks_usage_of_self_signed/

https://self-signed.badssl.com/ site returns a SEC_ERROR_BAD_DER.
I also have all my development virtual machines that are not accessible with Firefox 69.0 x64. I need to use Chrome instead.

Seems to be limited to Win10 Pro x64.

Comment 6

[Dana Keeler (she/her) (use needinfo) [:keeler]]

Can you attach a packet trace of the TLS handshake? Thanks! Also maybe try disabling antivirus "connection scanning" if you have it enabled?

Comment 7

[Cédric]

Attachment: badssl.pcapng

Here is a trace with self-signed.badssl.com with the new update 69.0.1

Comment 8

[Dana Keeler (she/her) (use needinfo) [:keeler]]

Thanks! Unfortunately there's nothing that stands out in that packet trace that would result in that error. Can you use https://mozilla.github.io/mozregression/quickstart.html to find when this stopped working?

Comment 9

[Stefan Matthäus]

Just tested Firefox 69.0.2 - Bug is still not fixed. Dell iDRAC web interface still not working. Still SEC_ERROR_BAD_DER error.

Get an old DELL server, try yourself.

Comment 10

[Cédric]

@Dana: It's not because someone did not have time to test your request that you can just close the ticket.
With a clean profile, I got no problem with mozregression but I still have the problem on 69.0.2.
May be I don't know how to work correctly with mozregression.
I'll retry as soon as I can with my profile as source.

@Stefan: If they did not mark it as confirm, it is because they don't reproduce it. You need not to be so condescending with a "Go buy a server!".

Comment 11

[Wayne]

We are seeing a similar issue with a self-signed certificate used by our internal application on Firefox 69. (Worked with prior versions.) When Firefox 69 tries to use the certificate SEC_ERROR_BAD_DER is displayed, and no option to accept the certificate.

Let me know if another data point would be useful, and I can provide specifics.

Comment 12

[Dana Keeler (she/her) (use needinfo) [:keeler]]

Wayne, Cédric, and Stefan, can you try with a recent nightly? (https://www.mozilla.org/en-US/firefox/channel/desktop/) This bug may have the same cause as bug 1570222, which was recently fixed.

Comment 13

[Stefan Matthäus]

I have checked in a virtual machine the recent nightly, it's not fixed. You can check by yourself by opening https://self-signed.badssl.com or the https idrac management interface of a Dell R610 server (this is the oldest machine I could test, I can see eBay.com US offers for working machine for just 49 bucks).

Please note, like said on above reddit posting, that in about:config the setting "security.enterprise_roots.enabled" must be set to "true" to reproduce the issue. This is default here in this company (I think, done through GPO), and the setting is greyed out in about:config, so it can't be changed by user. (I don't like to use Chrome instead of FF to go arround that issue.)

So you need to reopen that issue and investigate.

Comment 14

[Cédric]

I confirm Stefan assomption about "security.enterprise_roots.enabled" preference.
I removed it in my C:\Program Files\Mozilla Firefox\mozilla.cfg and I get now the self-signed error back.

Comment 15

[Dana Keeler (she/her) (use needinfo) [:keeler]]

With the enterprise roots feature enabled, can you please try:

• enabling the browser console by setting devtools.chrome.enabled to true in about:config
• opening the browser console (ctrl + shift + j)
• running Cc["@mozilla.org/psm;1"].getService(Ci.nsINSSComponent).getEnterpriseRoots() and Cc["@mozilla.org/psm;1"].getService(Ci.nsINSSComponent).getEnterpriseIntermediates()
• copying the output (best with right-click and Copy object)
• attaching the output to this bug in a text file
  Thanks!

Comment 16

[Cédric]

Attachment: enterprise.txt

Here, the requested command results.
Good luck !

Comment 17

[Dana Keeler (she/her) (use needinfo) [:keeler]]

Thanks! It appears the definitions mozilla::pkix is using for the tags for issuerUniqueID and subjectUniqueID (optional certificate fields that, evidently, aren't used much) have been incorrect since we shipped it 5 years ago (since the underlying type is BIT STRING, which is a "primitive" type, it shouldn't have the CONSTRUCTED bit set).

Comment 18

[Wayne]

I am unable to turn off "security.enterprise_roots.enabled" as it shows as "locked" when I do an about:config using the corporate customized version of Firefox. Is there another workaround until this gets fixed and released to an ESR?

Comment 19

[Dana Keeler (she/her) (use needinfo) [:keeler]]

Attachment: bug 1579060 - fix handling of issuerUniqueID and subjectUniqueID in mozilla::pkix::BackCert r?jcj

According to RFC 5280, the definitions of issuerUniqueID and subjectUniqueID in
TBSCertificate are as follows:

issuerUniqueID [1] IMPLICIT UniqueIdentifier OPTIONAL,
subjectUniqueID [2] IMPLICIT UniqueIdentifier OPTIONAL,

where UniqueIdentifier is a BIT STRING.

IMPLICIT tags replace the tag of the underlying type. For these fields, there is
no specified class (just a tag number within the class), and the underlying type
of BIT STRING is "primitive" (i.e. not constructed). Thus, the tags should be of
the form CONTEXT SPECIFIC | [number in class], which comes out to 0x81 and 0x82,
respectively.

When originally implemented, mozilla::pkix incorrectly required that the
CONSTRUCTED bit also be set for these fields. Consequently, the library would
reject any certificate that actually contained these fields. Evidently such
certificates are rare.

Comment 20

[J.C. Jones [:jcj] (he/they)]

checkin for NSS_3_47_BETA2

Comment 21

[J.C. Jones [:jcj] (he/they)]

Dana, can you check into the ASAN failures on try? https://treeherder.mozilla.org/#/jobs?repo=nss-try&revision=79e9a80dba89&selectedJob=270967832

Comment 22

[Dana Keeler (she/her) (use needinfo) [:keeler]]

Dealing with this in bug 1588567.

Comment 23

[J.C. Jones [:jcj] (he/they)]

https://hg.mozilla.org/projects/nss/rev/c50f933d37a5e35c916c1d52e1ee5cb1e19af730

Comment 24

[Stefan Matthäus]

Hello, I like to confirm, with Firefox 70, installed today, I am again able to open Dell iDRAC and https://self-signed.badssl.com/

Thanks for fixing!

Comment 25

[Cédric]

Thanks, also fixed for me.

Comment 26

[Wayne]

Confirmed our app is now working with Firefox 70.

Thanks!

Comment 27

[Ryan VanderMeulen [:RyanVM]]

Is this something likely to bite in an enterprise environment such that we should consider backporting to NSS 3.44 for ESR68 as well?

Comment 28

[Dana Keeler (she/her) (use needinfo) [:keeler]]

Yes - this is likely to affect enterprise users.

Comment 29

[J.C. Jones [:jcj] (he/they)]

NSS_3_44_BRANCH https://hg.mozilla.org/projects/nss/rev/64e55c9f658e2a75f0835d00a8a1cdc2f25c74d6

Comment 30

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/releases/mozilla-esr68/rev/ea1bc0fb2dda