[asterisk-bugs] [JIRA] (ASTERISK-29376) res_rtp_asterisk: Coredump with t.140 RED enabled

[Kevin Harwell (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-29376?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=255249#comment-255249 ] 

Kevin Harwell commented on ASTERISK-29376:
------------------------------------------

You've been running the above for a while and it hasn't crashed again?

If that's the case then given that a previous suggestion of
{quote}
if (rtp->red && rtp->red->t140.datalen > 0) {
{quote}
did not work (i.e. crashes still occurred), but your code above does then the only difference I see between the two is the check to see if {{rtp}} is NULL or not.

Is what you have the right fix? Not sure. It really depends on why, or how {{rtp}} is NULL at this point in time when it's retrieved.

If the {{rtp}} object is indeed the problem then one suggestion I have is to lock the instance prior to retrieving the {{rtp}} object. That will at least ensure it's not somehow freed between retrieving it and the NULL check.

Again though not saying it's the right fix. Someone would need to investigate more into why the object can be NULL here.

> res_rtp_asterisk: Coredump with t.140 RED enabled
> -------------------------------------------------
>
>                 Key: ASTERISK-29376
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-29376
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Resources/res_rtp_asterisk
>    Affects Versions: 17.8.1
>         Environment: RedHat 8
>            Reporter: Thomas Johnson
>            Assignee: Unassigned
>         Attachments: core.asterisk.2431.ipswitchdev66.microautomation.local.1617722782-brief.txt, core.asterisk.2431.ipswitchdev66.microautomation.local.1617722782-full.txt, core.asterisk.2431.ipswitchdev66.microautomation.local.1617722782-info.txt, core.asterisk.2431.ipswitchdev66.microautomation.local.1617722782-locks.txt, core.asterisk.2431.ipswitchdev66.microautomation.local.1617722782-thread1.txt, core.asterisk.296402.ipswitchdev66.microautomation.local.1622643818-brief.txt, debug_log_123456.gz, full, temp.zip
>
>
> Asterisk will randomly coredump if RED is allowed.  The exception is in an apparent scheduled callback to red_write.   I tried modifying the function to check for NULL pointers, but it still crashes due to possibly an invalid pointer.



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