Description of problem:
  The polkit daemon leaks processes zombie processes after timeout events. This results in a large number of Zombie processes building up and can result in system failures if not addressed.

Version-Release number of selected component (if applicable):
  polkit-0.112-14.el7.x86_64

How reproducible:
  Easily

Steps to Reproduce:
1. Add the following contents to a new /etc/polkit-1/rules.d/51-test.rules file

    polkit.addRule(function(action, subject) {          
        var res = polkit.spawn(['/usr/bin/sleep', '60']);                                                    
        polkit.log(res);      
        if (res == '')        
            return null;      
    });                       


2. Stop polkit

    # systemctl stop polkit

3. As a non-root user, issue a "systemctl restart test" which just triggers the polkit authentication configuration above.

    $ systemctl restart test


Actual results:
  The sleep processes are not reaped, leaving them to build up as observed in the following output:

    # ps --forest -p $(pidof polkitd) --ppid $(pidof polkitd)| head
      PID TTY          TIME CMD
      621 ?        00:00:03 polkitd
     3351 ?        00:00:00  \_ sleep <defunct>
     3468 ?        00:00:00  \_ sleep <defunct>
     3494 ?        00:00:00  \_ sleep <defunct>
     3520 ?        00:00:00  \_ sleep <defunct>
     3546 ?        00:00:00  \_ sleep <defunct>
     3572 ?        00:00:00  \_ sleep <defunct>
     3598 ?        00:00:00  \_ sleep <defunct>
     3625 ?        00:00:00  \_ sleep <defunct>



Expected results:
  Zombie sleep processes are properly reaped.


Additional info:
  This issue is observable in the latest Fedora releases and has been reported in the upstream "PolicyKit" project bug tracker:

    106021 – Polkitd: The utils_spawn_data_free reap timeout subprocess did not work resulting in a large number of zombie processes 
    https://bugs.freedesktop.org/show_bug.cgi?id=106021