[
https://issues.apache.org/jira/browse/TS-4480?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15316738#comment-15316738
]
James Peach commented on TS-4480:
---------------------------------
I thought about this a bit, and I wonder why we even need to do a trie search
here. Originally we used a trie because we wanted to do a longest match. If we
are not doing a longest match, then for a name "foo.example.com", we could just
do hash lookups for "foo.example.com" and "example.com". One of the original
ideas about using a trie was that we keep it as the single index, but it ended
up being too slow and memory intensive to use for large numbers of certificates.
> Wildcards in certificates should only match one level
> -----------------------------------------------------
>
> Key: TS-4480
> URL: https://issues.apache.org/jira/browse/TS-4480
> Project: Traffic Server
> Issue Type: Bug
> Components: Core, SSL
> Reporter: Michael Sokolnicki
> Fix For: 7.0.0
>
> Attachments: current_patch.diff
>
>
> According to RFC 6125 section 6.4.3:
> {quote}
> If the wildcard character is the only character of the left-most label in the
> presented identifier, the client SHOULD NOT compare against anything but the
> left-most label of the reference identifier (e.g., *.example.com would match
> foo.example.com but not bar.foo.example.com or example.com).
> {quote}
> In the current implementation, certificates are searched for in a trie, and
> the longest match is returned, but there is no check if that match complies
> with the above rule. This causes invalid certs to be returned and SLL errors
> in the browser (in Firefox, we get SSL_ERROR_BAD_CERT_DOMAIN).
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
