[asterisk-bugs] [JIRA] (ASTERISK-27862) ARI: Crashing on json_mem_free (after sending event)

Stefan Repke (JIRA) noreply at issues.asterisk.org
Wed May 23 04:06:56 CDT 2018


    [ https://issues.asterisk.org/jira/browse/ASTERISK-27862?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=243507#comment-243507 ] 

Stefan Repke commented on ASTERISK-27862:
-----------------------------------------

We use the Debian package libjansson of version 2.7-1+deb8u1.

Further, I was curious about the advantages you mentioned and moved to Debian package 2.11-1. Then, all of the above problems are gone:
* no crashes!
* no "Memory corrupted after free of 0x... allocated at json.c ast_json_malloc() line 155"!
* no "ERROR\[...\]: ari/ari_websockets.c:175 ast_ari_websocket_session_write: Failed to encode JSON object"!

Hence, Asterisk has threadding issues when using an older version of libjansson (like Debian stable) ...

> ARI: Crashing on json_mem_free (after sending event)
> ----------------------------------------------------
>
>                 Key: ASTERISK-27862
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-27862
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Applications/app_stasis
>    Affects Versions: 13.21.0
>            Reporter: Stefan Repke
>            Assignee: Stefan Repke
>         Attachments: backtrace 1.txt, backtrace 2.txt, backtrace 3.txt, cli.txt
>
>
> ARI is activated and there are several apps connected with "subscribeAll=true".
> Then, Asterisk crashes occasionally when delivering ARI events!
> With normal workload, this happens once or twice a day. It can be provoked by adding many (eg. 10) ARI subscribers and producing lots of events (eg. PeerStatusChange by registering/unregistering a peer).
> I coredumped three crashes, all having a memory corruption in {{json_mem_free}} at {{json.c:84}}; cf. attached backtraces.
> * Backtrace 2 and 3 have similar problems: {{double free or corruption (fasttop)}},
> * Backtrace 1 looks like wrong memory access.
> I've also attached the last logger messages from the CLI (debug=5 / verbose=5).



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



More information about the asterisk-bugs mailing list