Re: initscripts proposal conflict?

[Thorsten Kukuk <kukuk@suse.de>]

Hi,

On Wed, Jul 19, Guilherme Manika wrote:

> 
>   The init scripts proposal seems to be self-conflicting to
> me. sysinit/initactions.sgml says:
> 
> "Init files should ensure that they will behave sensibly if invoked
> with start when the service is already running, or with stop when it
> isn't, and that they don't kill unfortunately-named user
> processes. (...)"
> 
>   sysinit/initfunctions.sgml defines a algorithm for pidofproc and
> killproc that does not seem to comply with this. If the basename.pid
> file is not available, it uses pidof to find the running process.
> However, pidof(8)'s man page says that it is not safe if all it gets
> is the basename of the process (it may return unfortunately-named user
> processes otherwise). The usage of ps in case pidof fails is also
> broken in the same sense.
> 
>   Possible solutions: either require that killproc and pidofproc get a
> complete path as an argument, and usage of 'ps' as one of the possible
> ways to get the pid of a process is dropped; or define that
> start-stop-daemon should be present and use it (I believe s-s-d
> doesn't have the problems described).

I think killproc and pidofproc should get the complete path as an
argument. We do it for SuSE Linux since a long time, it is the only
secure way not to kill the wrong process.

  Thorsten

-- 
Thorsten Kukuk       http://www.suse.de/~kukuk/       kukuk@suse.de
SuSE GmbH            Schanzaeckerstr. 10            90443 Nuernberg
Linux is like a Vorlon.  It is incredibly powerful, gives terse,
cryptic answers and has a lot of things going on in the background.