[asterisk-bugs] [JIRA] (ASTERISK-20947) astcanary exits immediately because of wrong pid argument

[Richard Mudgett (JIRA)]

     [ https://issues.asterisk.org/jira/browse/ASTERISK-20947?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Richard Mudgett updated ASTERISK-20947:
---------------------------------------

    Description: 
I wanted to update our machines to 10.12.0 (as an intermediate update
until we upgrade to 11) and noticed that astcanary is not running any
more, not even right after starting asterisk, resulting in asterisk to
reduce it's priority (with a rather lengthy and slightly annoying
message). This only happens when starting asterisk in the background,
though.

With strace I found out that astcanary being started with a wrong ppid:

{noformat}
> 29358 18:19:04.138884 execve("/usr/sbin/astcanary", ["astcanary", "/var/run/asterisk/alt.asterisk.c"..., "29356"], [/* 28 vars */] <unfinished ...>
...
> 29358 18:19:04.146543 setpriority(PRIO_PROCESS, 0, 0 <unfinished ...>
> 29358 18:19:04.146568 <... setpriority resumed> ) = 0
> 29358 18:19:04.146593 getppid( <unfinished ...>
> 29358 18:19:04.146633 <... getppid resumed> ) = 29357
> 29358 18:19:04.146663 exit_group(0)     = ?
{noformat}

Notice 29356 vs. 29357.

The reason is obviously a change made to main/asterisk.c. In 10.11.1,
ast_mainpid is updated right after daemon(), in 10.12.0 it is updated
_after_ astcanary is started, so astcanary gets the old pid (the process
starting doing the daemon()) and therefore exits immediately because it checks the ppid argument against the return value of getppid().

The easy fix is to update ast_mainpid after daemon(), as in the attached patch.

  was:
I wanted to update our machines to 10.12.0 (as an intermediate update
until we upgrade to 11) and noticed that astcanary is not running any
more, not even right after starting asterisk, resulting in asterisk to
reduce it's priority (with a rather lengthy and slightly annoying
message). This only happens when starting asterisk in the background,
though.

With strace I found out that astcanary being started with a wrong ppid:

> 29358 18:19:04.138884 execve("/usr/sbin/astcanary", ["astcanary", "/var/run/asterisk/alt.asterisk.c"..., "29356"], [/* 28 vars */] <unfinished ...>
...
> 29358 18:19:04.146543 setpriority(PRIO_PROCESS, 0, 0 <unfinished ...>
> 29358 18:19:04.146568 <... setpriority resumed> ) = 0
> 29358 18:19:04.146593 getppid( <unfinished ...>
> 29358 18:19:04.146633 <... getppid resumed> ) = 29357
> 29358 18:19:04.146663 exit_group(0)     = ?

Notice 29356 vs. 29357.

The reason is obviously a change made to main/asterisk.c. In 10.11.1,
ast_mainpid is updated right after daemon(), in 10.12.0 it is updated
_after_ astcanary is started, so astcanary gets the old pid (the process
starting doing the daemon()) and therefore exits immediately because it checks the ppid argument against the return value of getppid().

The easy fix is to update ast_mainpid after daemon(), as in the attached patch.

    
> astcanary exits immediately because of wrong pid argument
> ---------------------------------------------------------
>
>                 Key: ASTERISK-20947
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-20947
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>          Components: Utilities/astcanary
>    Affects Versions: 10.12.0
>         Environment: all that HAVE_WORKING_FORK
>            Reporter: Jakob Hirsch
>         Attachments: asterisk-10.12.0.astcanary_ppid.diff
>
>
> I wanted to update our machines to 10.12.0 (as an intermediate update
> until we upgrade to 11) and noticed that astcanary is not running any
> more, not even right after starting asterisk, resulting in asterisk to
> reduce it's priority (with a rather lengthy and slightly annoying
> message). This only happens when starting asterisk in the background,
> though.
> With strace I found out that astcanary being started with a wrong ppid:
> {noformat}
> > 29358 18:19:04.138884 execve("/usr/sbin/astcanary", ["astcanary", "/var/run/asterisk/alt.asterisk.c"..., "29356"], [/* 28 vars */] <unfinished ...>
> ...
> > 29358 18:19:04.146543 setpriority(PRIO_PROCESS, 0, 0 <unfinished ...>
> > 29358 18:19:04.146568 <... setpriority resumed> ) = 0
> > 29358 18:19:04.146593 getppid( <unfinished ...>
> > 29358 18:19:04.146633 <... getppid resumed> ) = 29357
> > 29358 18:19:04.146663 exit_group(0)     = ?
> {noformat}
> Notice 29356 vs. 29357.
> The reason is obviously a change made to main/asterisk.c. In 10.11.1,
> ast_mainpid is updated right after daemon(), in 10.12.0 it is updated
> _after_ astcanary is started, so astcanary gets the old pid (the process
> starting doing the daemon()) and therefore exits immediately because it checks the ppid argument against the return value of getppid().
> The easy fix is to update ast_mainpid after daemon(), as in the attached patch.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira