On 08/28/2011 07:08 PM, Ludovic Rousseau wrote:
> Hello Kalev,
> 
> I am revewing your systemd patches.

Hello Ludovic,


> Is this patch functionally needed or is it just to remove dead code 
> (when systemd is used)?

I wanted to make sure that when using Fedora packages, the code for
autostarting pcscd by forking from user space library was completely
disabled. It just doesn't work properly with SELinux.

See e.g. https://bugzilla.redhat.com/show_bug.cgi?id=639373 - "SELinux
is preventing /usr/lib/nspluginwrapper/npviewer.bin "execute" access on
/usr/sbin/pcscd."

Even though the Fedora package enables systemd socket activation by
default, a system administrator could possibly disable the pcscd service
/ socket. When the service is disabled, I would expect it to stay
disabled. However, without this patch, libpcsclite would fall back to
executing pcscd, filling the system system logs with SELinux denials.


> If I am correct systemd will create the socket itself (before pcscd 
> is started). So the function SCardCheckDaemonAvailability() will 
> return SCARD_S_SUCCESS and the client library should not fork and 
> exec the daemon. Am I right?

Yes, the patch shouldn't have any effect when systemd has created the
socket and passed it to pcscd. The patch makes sure that when systemd
socket activation is _disabled_, pcscd wouldn't fall back to the old
startup code.

I can imagine some other distributions, e.g. Debian that don't use
SELinux by default and don't use systemd by default either, would
probably want to keep the old autostart code enabled, but also compile
in (optional) systemd autostart support.

The compile time option makes it possible to support both use cases.


Thanks,
Kalev
_______________________________________________
Muscle mailing list
[email protected]
http://lists.drizzle.com/mailman/listinfo/muscle

Reply via email to