[asterisk-bugs] [JIRA] (ASTERISK-29436) core/channel : DTMF emulation doesn't queue the end when last RTP received is the END frame

[Arnaud Willem (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-29436?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=254981#comment-254981 ] 

Arnaud Willem commented on ASTERISK-29436:
------------------------------------------

Hello,

Will test against a fresh compile of the latest Asterisk 16 and 18 branches and provide feedback.

For reference, this commit fixes a similar issue (ongoing DTMF regen/emulation) but with a different cause (vanishing channel from the bridge) : 6ad0126425993bd9c9c73b0c05a52984cbcadc5d

> core/channel : DTMF emulation doesn't queue the end when last RTP received is the END frame
> -------------------------------------------------------------------------------------------
>
>                 Key: ASTERISK-29436
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-29436
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Core/Channels
>    Affects Versions: 16.2.1
>         Environment: Debian/stable - Xen HVM - Linux 4.19
>            Reporter: Arnaud Willem
>            Assignee: Arnaud Willem
>            Severity: Major
>         Attachments: ast_bugreport.txt
>
>
> In this specific context, a PJSIP peer (peerA) is sending RFC4733 DTMF towards Asterisk.
> Asterisk then forwards it to the PSTN over chan_pjsip (peerB).
> peerA sends sequences of 2-3 DTMF digits. When the last digit is sent, it stops sending RTP frames and awaits responses from the called party.
> This appears to trigger an odd "off-by-one" bug in main/channel.c:__ast_read() where an additional frame is expected to stop DTMF emulation in the queue.
> Attached log shows this happening twice, around lines 92 (begin emulation of DTMF 8 to peerB ... but no END until line 257 where a new rfc4733 frame has previously been received from peerA.) and 286, (begin emulation of DTMF C to peerB, then the call is hung up after a while.)
> A pcap dump can also be provided if necessary.



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