LVS
lvs-devel
Google
 
Web LinuxVirtualServer.org

Re: unregister_netdevice: waiting for lo to become free. Usage count = 8

To: Hans Schillstrom <hans@xxxxxxxxxxxxxxx>
Subject: Re: unregister_netdevice: waiting for lo to become free. Usage count = 8
Cc: Simon Horman <horms@xxxxxxxxxxxx>, netdev@xxxxxxxxxxxxxxx, lvs-devel@xxxxxxxxxxxxxxx, "Eric W. Biederman" <ebiederm@xxxxxxxxxxxx>
From: Julian Anastasov <ja@xxxxxx>
Date: Tue, 19 Apr 2011 00:12:27 +0300 (EEST)
        Hello,

On Mon, 18 Apr 2011, Hans Schillstrom wrote:

> Actually I forgot to tell there is a need for a
> ip_vs_service_cleanup() due to above.
> Do you see any drawbacks with it ?

        May be ip_vs_service_cleanup() should call only
ip_vs_flush(), under __ip_vs_mutex.

> /*
>  *    Delete service by {netns} in the service table.
>  */
> static void ip_vs_service_cleanup(struct net *net)
> {
>       unsigned hash;
>       struct ip_vs_service *svc, *tmp;
> 
>       EnterFunction(2);
>       /* Check for "full" addressed entries */
>       for (hash = 0; hash<IP_VS_SVC_TAB_SIZE; hash++) {
>               write_lock_bh(&__ip_vs_svc_lock);
>               list_for_each_entry_safe(svc, tmp, &ip_vs_svc_table[hash],
>                                        s_list) {
>                       if (net_eq(svc->net, net)) {
>                               ip_vs_svc_unhash(svc);
>                               __ip_vs_del_service(svc);
>                       }
>               }
>               list_for_each_entry_safe(svc, tmp, &ip_vs_svc_fwm_table[hash],
>                                        f_list) {
>                       if (net_eq(svc->net, net)) {
>                               ip_vs_svc_unhash(svc);
>                               __ip_vs_del_service(svc);
>                       }
>               }
>               write_unlock_bh(&__ip_vs_svc_lock);
>       }
>       LeaveFunction(2);
> }
> 
> Called just after the __ip_vs_control_cleanup_sysctl()

        Hm, no. ip_vs_service_cleanup() should be called
by ip_vs_cleanup() before or after nf_unregister_hooks().
The rule is that ip_vs_flush() should be called before
ip_vs_conn_flush() because after ip_vs_flush() no more
connections can be created and even if hooks are still
registered the packets can not create conns in the netns. Then
ip_vs_conn_flush() will remove all existing connections and
ip_vs_control_cleanup() can remove all real servers with
ip_vs_trash_cleanup(). I mean, per-netns calls.

        Also, may be all code that was called in old
kernels by ip_vs_cleanup() should be now called by
__ip_vs_cleanup(net), i.e. we can preserve the needed order
of all functions but now also per-netns. For example, for
ip_vs_ctl.c ip_vs_control_init() can remain as global but it
should not register ipvs_control_ops. Then we
can rename __ip_vs_control_init to ip_vs_control_init_net()
and to call it from __ip_vs_init(). I.e. all such files
will have global function and also _init_net and
_cleanup_net. Now there are many register_pernet_subsys()
calls and I'm not sure we preserve the needed order for
cleanup. Are the ->exit methods called in reverse order?
I don't see it in ops_exit_list() and we can not rely
on such registration order. I think, ip_vs_init() should
call global functions as now but __ip_vs_init() and
__ip_vs_cleanup() should call the _net methods in right
order.

Regards

--
Julian Anastasov <ja@xxxxxx>
--
To unsubscribe from this list: send the line "unsubscribe lvs-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html

<Prev in Thread] Current Thread [Next in Thread>