[asterisk-bugs] [JIRA] (ASTERISK-28798) [patch] chan_sip: TCP/TLS client without server.

Kevin Harwell (JIRA) noreply at issues.asterisk.org
Fri Jun 26 09:39:25 CDT 2020


    [ https://issues.asterisk.org/jira/browse/ASTERISK-28798?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=251258#comment-251258 ] 

Kevin Harwell commented on ASTERISK-28798:
------------------------------------------

[~traud]
{quote}
Digium has a hard to understand (or at least an unexplained) policy when it comes to issue tracking.
{quote}
>From a software perspective while an issue may be related, or even caused by another they are separate issues. The original issue here has been fixed, and marked as such. Any additional work, or related problems should get a new issue in JIRA, but linked back to this one.

>From a business/planning perspective it makes things easier to track and helps to break things into smaller "parts" in order to better evaluate a particular issue's business impact, and engineering effort.

Having a separate issue for each problem, and potentially subsequent patch also helps to track each individual issue from initial report, to review and commit, and into a release. A side effect of that also, yes, helps to script reporting. It also allows us to easier keep up with any associated attachments, and other related documents.

Overall you can think of it similar to the idea of "separation of concerns", and breaking up software into smaller easier to maintain pieces. Sure I can have one file, and one main function for my program but that's going to make things much harder in the long run.

I hope that helps some.

> [patch] chan_sip: TCP/TLS client without server.
> ------------------------------------------------
>
>                 Key: ASTERISK-28798
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-28798
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Channels/chan_sip/TCP-TLS
>    Affects Versions: 13.32.0, 16.9.0, 17.3.0
>            Reporter: Alexander Traud
>            Assignee: Alexander Traud
>            Severity: Minor
>              Labels: patch
>      Target Release: 13.33.0, 16.10.0, 17.4.0
>
>         Attachments: client_without_server.patch
>
>
> The channel driver {{chan_sip}} can be configured for various SIP transports like UDP, TCP, and TLS. Because it is a back-to-back user agent (B2BUA), it can be configured as client and/or server. For example, when you add just
> {code}register => tcp://user:secret@host{code}
> to the {{\[general\]}} context in the configuration file {{sip.conf}} (just that, nothing else; the file is empty), a client connection to a remote VoIP/SIP server like a PSTN service is established. However in the SIP-REGISTER message, the headers Via and Contact do not contain the local IP address but the value {{(null)}}. Such a value is rejected by a remote party normally.
> Currently, the source code expects that the local server also has TCP enabled. Consequently, for a client connect you have to write at least:
> {code}tcpenable=yes
> register => tcp://user:secret@host{code}
> In case of TLS, something like
> {code}
> tlscapath=/etc/ssl/certs/
> register => tls://user:secret@host
> {code}should be sufficient. However, that gives the same, the value {{(null)}} in the SIP-REGISTER for the Via and Contact headers. Again, the TLS server must be enabled locally which requires a public certificate with private key:
> {code}
> tlsenable=yes
> tlscertfile=asterisk.pem
> tlscapath=/etc/ssl/certs/
> register => tls://user:secret@host
> {code}Consequently, I have to create a (server) certificate locally, do create a remote TLS client connection successfully. This is counter-intuitive. The situation is even worse because there is no log message and the SIP-REGISTER is sent with wrong information. Only at debug level 3, one can see that the local bind address is null.
> The cause for this is the function {{ast_sip_ouraddrfor(.)}} which does not check whether the {{local_address}} is null. See the attached patch.
> However, there are two more places in code which do not check for null:
> 1. function {{get_address_family_filter(.)}}
> 2. {{p->socket.port}} in function {{transmit_register(.)}}
> I have no idea what to do with those.



--
This message was sent by Atlassian JIRA
(v6.2#6252)



More information about the asterisk-bugs mailing list